Policy Research Working Paper 10910 Using Poverty Lines to Measure Refugee Self-Reliance Johannes Hoogeveen Robert Hopper Poverty and Equity Global Practice September 2024 Policy Research Working Paper 10910 Abstract Humanitarian models of refugee assistance increasingly countries, while refugees residing in urban and non-camp promote refugee self-reliance without offering a clear under- settings demonstrate higher levels of self-reliance than those standing of what constitutes self-reliance, or how to measure in rural and camp environments, reflecting the presence it, although measurement is essential to assess whether the of greater economic opportunities outside camps and in promotion of self-reliance has been successful. This paper urban areas. The paper also finds evidence that aid and proposes an approach to measuring self-reliance rooted in self-reliance are inversely correlated, with more aid spent global poverty measurement: a refugee is self-reliant if their in environments where refugee self-reliance is constrained. self-earned income exceeds the locally relevant poverty line. The paper argues that a different outcome should be envis- In its empirical application, the paper uses estimates of aged, one in which governments that encourage refugees to self-reliance drawn from 11 consumption surveys, which be self-reliant receive more aid. Such an approach would be present the universe of data that can be used for such pur- beneficial for refugees, who would gain financial autonomy, poses. Refugees in middle-income countries are found to be and for host country citizens, who often face high levels of far more likely to be self-reliant than those in lower income poverty and at times are even poorer than refugees. This paper is a product of the Poverty and Equity Global Practice. 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 jhoogeveen@worldbank.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 Using Poverty Lines to Measure Refugee Self-Reliance 1 Johannes Hoogeveen and Robert Hopper JEL Classification: I32 Measurement and Analysis of Poverty; O12 Microeconomic Analyses of Economic Development; O15 Human Resources; Human Development; Income Distribution; Migration. 1 Microdata used in this paper to estimate refugee income were provided by: Agustin Arakaki, Aziz Atamanov, Juan Monroy Barragan, Mohamed Coulibaly, Maria Davalos, Alvin Etang, Arden Finn, Sebastian Leander, Vincenzo di Maro, Chinedu Obi, Aboudrahyme Savadogo, Ganesh Seshan, Nistha Sinha, Wondimagegn Tesfaye and Christina Wieser. Suggestions for improvement received from Johan Mistiaen are gratefully acknowledged. 1. Introduction Today, more than 36 million refugees are displaced worldwide (UNHCR 2024a). Of these, an estimated 67 percent possess full legal rights to labor market participation, though in practice, only 45 percent are estimated to have unrestricted access (UNHCR 2023). With many refugees facing restrictions to their freedom of movement and right to work, dependency on humanitarian aid is commonplace among refugees, inhibiting their self-reliance. With durable solutions remaining elusive, the long-standing default model of refugee assistance – one that supplants socio-economic rights with external assistance – is increasingly being seen as outdated and unsustainable, with support rapidly moving away from the traditional approach to refugee assistance, to more sustainable and progressive programs and policies that actively seek to promote self-reliance (Betts 2021). While traditional, ‘care and maintenance’ forms of humanitarian assistance, such as the provision of food and shelter, are often critical at the onset of refugee crises, they become increasingly problematic when they are sustained over the longer term. As emergency phases subside, there is increasing recognition that traditional types of humanitarian aid should play a more limited role in refugees’ lives, with support instead directed towards job creation and building the economic capacities of refugees (Easton-Calabria and Omata 2018). The international discourse has consequently shifted its focus away from long-term, humanitarian models of assistance towards the promotion of refugee self-reliance (Betts 2021) and sustainable programming (UNHCR 2024b). Yet despite this consensus, there remains a clear lack of understanding around what constitutes self-reliance, and by extension, how to measure it. Without measurement, it is not only difficult to understand whether the promotion of self-reliance has been successful, but almost impossible to nurture the systemic learning required to achieve refugee self- reliance in the long-term. To date, only a handful of approaches have been proposed to measure refugee self-reliance, each of which comes with its own limitations. These measures are either at odds with the concept of self-reliance – in that refugees who do depend on humanitarian assistance can be considered self-reliant – or where a proposed measure is consistent, the data required to measure it is not (yet) available (e.g. the measure proposed by Hoogeveen and Hopper, 2024). In this paper, the aim is to measure self-reliance using a simple one-dimensional indicator: income. Here, a refugee is considered self-reliant if their self-earned income exceeds the poverty line relevant to their context. It is thus not sufficient for refugees to survive on humanitarian assistance; self-reliance instead requires that refugees have the ability to be independent of assistance and generate the income necessary to attain a – locally relevant – acceptable minimum standard of living. This approach has several advantages. Practically, it anchors refugee self-reliance within existing poverty measurement, including the monitoring of SDG1.1 (elimination of extreme poverty). Empirically, it allows one to explore the determinants of refugee self-reliance through a quantifiable and widely understood metric: income generation. Theoretically, it offers an appropriate measure with which to conceptualize refugee needs, capturing both income and, at least partially, non-income factors in its measurement, as this paper will demonstrate, using estimates of self-reliance drawn from 11 consumption surveys, which present the universe of data that can be used for such purposes. As far as we can tell, this paper is the first to present a set of global descriptives on refugee self-reliance. The remainder of the paper is organized as follows. The next section reviews the literature on, and current measurements of, self-reliance. Section 3 outlines the paper’s proposed measure self-reliance and 2 approach to the existing data on refugee self-reliance using 11 microdata surveys from low- and middle- income countries. Section 4 discusses the results and the potential implications of the findings. Section 5 draws key conclusions. 2. Review of literature on measuring self-reliance The most widely used definition of refugee self-reliance comes from the United Nations High Commissioner for Refugees (UNHCR, 2014) which defines self-reliance as “the ability of an individual, household or community to meet essential needs and to enjoy social and economic rights in a sustainable manner and with dignity. By becoming self-reliant, refugees and displaced persons lead active and productive lives and are able to build strong social, economic and cultural ties with their host communities. Self-reliance can assist in ensuring that persons of concern are better protected by strengthening their capacity to claim their civil, cultural, economic, political and social rights” (UNHCR, 2014). Self-Reliance has emerged as a priority outcome of refugee support in recent years. Driven by landmark agreements of the New York Declaration for Refugees and Migrants (2016), its accompanying Comprehensive Refugee Response Framework (CRRF), and the UN Global Compact on Refugees (2018), there is more emphasis than ever on the promotion of refugee self-reliance (Betts, 2021; Easton-Calabria, 2021; UNHCR 2024b). However, the centrality of self-reliance to international refugee discourse marks more of a resurgence than an innovation, as self-reliance has been a repeated goal of assistance actors for over a century (Easton-Calabria, 2021). Since its emergence in 1920s Greece, when the League of Nations provided livelihood assistance to over 1.5 million refugees, self-reliance has been through many different incarnations, all of which have repackaged similar practices under different monikers, such as ‘self- sufficiency’, ‘self-supporting’, ‘rehabilitation’, ‘self-help’, ‘rural animation’ and ‘sustainable programming’ (Betts, 2021; Easton-Calabria, 2022; Easton-Calabria and Omata, 2018; UNHCR 2024b). Despite refugee self-reliance, in its various forms, existing for over a century, self-reliance has rarely been measured, and in the rare instances where it has, these measures have been deficient, failing to capture the central tenet of refugee self-reliance: independence from aid (Betts, 2021; Schön, 2020; Hoogeveen and Hopper, 2024). The most cited measures of refugee self-reliance are the ‘Self-Reliance Index’ (SRI), developed by the Refugee Self-Reliance Initiative who proclaim their measure to be “the first global tool to measure the progress of refugee families on their journey to self-reliance" (SRI, n.d.)., and Betts et al.’s (2018, 2020) measure of refugee self-reliance, which is used to assess how well-adapted refugee settlements are to supporting refugees. Despite representing the pre-eminent (if not only) recognized measures on refugee self-reliance, neither measure adequately accounts for independence from aid, thereby failing to capture the true notion of ‘self-reliance’. For instance, both the SRI and Betts et al. (2018, 2020) account for food security in their measures, but do so by measuring food consumption only, with no consideration given to the source of this food consumption and whether food security needs can be met independently of aid. While both measures include information on independence from aid, this information does not inform the other metrics used by the authors to calculate self-reliance, with refugees deemed to be ‘self-reliant’ in food security, public goods, health, education and other domains once a certain outcome is achieved, regardless of whether these outcomes can be achieved in the absence of external assistance. The benefits and shortcomings of these measures are discussed in greater detail in Hoogeveen and Hopper (2024), who present their own, multi-dimensional measure of refugee self-reliance, while this paper offers a more simplified measure of self-reliance, one that is achievable given current limitations in the data. The 3 authors recognize that this indicator does not incorporate all aspects of self-reliance, but by focusing on self-earned income, it arguably captures the most critical aspects of self-reliance: the ability to meet ones physiological and basic material needs independently of aid. In fact, given the challenging environments in which many refugees live, and the primacy attached to physiological and material needs, ensuring that these minimum, essential needs are met is of overriding importance, and this focus could be diluted by accommodating multiple other objectives when measuring of self-reliance (Greeley, 1994). This measure will not be satisfactory to some readers, especially as refugees face major constraints to both income generation as well as access to services and basic entitlements, constraints which are not imposed on host citizens. For these persons, the concept of refugee self-reliance should inevitably be broadened to capture these additional constraints, yet there are two responses to this observation. On the one hand, if labor market restrictions and unequal treatment exceed a certain threshold, refugees will not be able to earn incomes that exceed the poverty line and self-reliance will not be attained. Similarly, if refugees have no access to education or health care, their income-generating capacity will be undermined and this will be reflected in the incomes they earn. As such, this paper’s proposed self-reliance indicator not only captures a minimum acceptable standard of living, but implicitly captures how much discrimination and exclusion refugees experience through the ‘relative’ component of poverty captured by the global poverty lines. This reasoning will not appeal to all, and just as global welfare measurements have expanded to report not only monetary poverty but multi-dimensional poverty (of which monetary poverty is an important element), one could develop a multi-dimensional self-reliance indicator. This measure would likely capture the monetary self-reliance indicator proposed in this paper, along with indicators on labor market regulations (de jure and de facto), access to services, and other key factors. Such a metric is explored further in Hoogeveen and Hopper (2024), who present a multi-dimensional measure of self-reliance that expands on the approach outlined in this paper, while addressing the shortcomings of the SRI and Betts et al.’s indicators. 2 3. Approach: The International Poverty line as a minimum standard for self-reliance This section explores how to define “the ability to meet essential needs”, the central tenet of this paper’s definition, and measurement, of refugee self-reliance. The necessity to identify a minimum acceptable standard is not unique to self-reliance. Welfare analytics of the most vulnerable in society faced a similar challenge in the 1990s, which it resolved by defining a ‘poverty line’. Those consuming less than the poverty line are poor, those consuming more are non-poor. 2 Their multidimensional indicator covers five key areas: (i) the ability of refugees to meet their essential needs (exceed a locally relevant poverty line) through self-earned income; (ii) whether refugees have the right to access public schools and their level of school enrollment; (iii) whether refugees have the right to access public health care combined with their access to health care; (iv) whether refugees have freedom of movement, and if they do, whether they can move without harassment; and (v) whether refugees have a right to water in accordance with local legislation, and whether this is achieved in practice. This indicator thus establishes whether refugees are able to earn a sufficient income and access critical public services without humanitarian assistance, while simultaneously achieving certain outcome levels in these areas. The major challenge with this indicator is that no data set currently exists that measures refugee self-reliance in all five of these dimensions sufficiently. 4 Poverty lines have been widely adopted and can be set in different ways. Typically, they are anchored to nutritional requirements - a minimum caloric amount - and then complemented with the cost to satisfy non-food essentials (clothing, shelter, private expenditure on education and health, and so on) obtained from households whose total (food) consumption equals the cost of obtaining the nutritional requirements. Poverty lines are typically derived using data from consumption surveys and as such reflect the actual consumption and revealed preferences of households. Poverty lines are thus context-specific, reflecting the consumption needs for achieving the minimum standard in a given society (Ravallion et al. 2008). By implication, each country sets its own national poverty line. Global poverty measurement does not use national poverty lines, but uses the International Poverty Line (IPL), which itself is anchored to the national poverty lines of the world’s poorest countries. This first international poverty line, based on research by Ravallion, Datt, and van de Walle (1991), was set a PPP$ 1.02 per day per person and became known as the dollar-a-day poverty line. Since the early 1990s the IPL has been updated with some regularity. The latest update was produced in 2022 following the release of the 2017 PPPs. In this iteration, the IPL was defined as the median of the national poverty lines of 28 of the world’s poorest countries and set at PPP$ 2.15, with SDG 1 using it as a benchmark for poverty eradication (Joliffe et al. 2022). With the 2017 PPP update came the recognition that the IPL may be too low to act as an acceptable minimum standard of living for middle-income countries, with subsequent global poverty lines for lower-middle, upper-middle and even high-income countries published thereafter. In the remainder of this paper, global poverty lines are used to explore self-reliance, with refugees deemed self-reliant when they earn an income greater than the applicable global poverty line. = 1 if > , 0. In the equation above, si is the self-reliance indicator of refugee i, z is the poverty line, and yi is earned income – defined as the income from wage employment and earnings from self-employment. Earned income does not include gifts (either in-kind or cash) from humanitarian agencies and other sources, nor does it include remittances from non-refugee family members. The self-reliance indicator can be expressed as the average incidence of self-reliance among the refugee population as follows: 1 − 0 = �� � z =1 Here, n is the number of refugees and q the number of refugees that are self-reliant. Note that for refugees to be self-reliant, their income yi has to exceed the poverty line z so that the bracketed term equals 1 for self-reliant refugees. This is, of course, a complex way of writing down the fraction of self-reliant people, but doing so has the advantage of clarifying how the self-reliance indicator S relates to the Foster–Greer– Thorbecke (FGT) family of poverty metrics (Foster et al. 1984), defined as: 1 − = � � � z =1 5 where n is the number of refugees and p is the number of refugees that are not self-reliant (n = p + q). When equals zero (poverty incidence) it holds that FGT0 + S = 1: the fraction of poor refugees plus the fraction of self-reliant refugees equals one as a refugee is either poor, or self-reliant. Note that there are at least two critical differences with traditional poverty measurement. First, self-reliance is derived from earned income, whereas poverty is typically, but not always, measured through consumption. Being able to sustain oneself through self-earned income is critical to a dignified life, with an inability to do so linked to "psychological harm, loss of work, skill and self-confidence, increase in ailments and morbidity, disruption of family relations and social life, hardening of social exclusion, and accentuation of racial tensions and gender asymmetries" (Sen 1999, p.94). This is not to state that all work is good, as there exist tyrannical forms of work that are major sources of deprivation. Yet as self-earned income is essential to self-reliance, and with this caveat in mind, a refugee is classified as self-reliant when their self-earned income exceeds the poverty line; while a refugee is considered non-poor when their consumption exceeds the poverty line. As such, a refugee who does not work and who is sustained by humanitarian assistance cannot be self-reliant. While a refugee is self-reliant if their earned income exceeds the poverty line, but may be classified as poor if the consumption from this income is less than the poverty line. This occurs, for instance, if part of the income is saved or invested. 3 Second, the emphasis in this paper’s measure of self-reliance is on self-earned income. Gifts and humanitarian assistance are a source of income but are not self-earned. The same holds for income from remittances. Poverty measurements, by contrast, include consumption derived from gifts, or items purchased from remittances or on credit. Provided only refugee self-earned income is considered, FGT1, also known as the poverty gap, can be interpreted as the ‘self-reliance gap’, which measures by how much, expressed as a fraction of the poverty line, refugee incomes would have to rise to make all refugees self-reliant. Measuring income precisely through household surveys is notoriously difficult (Deaton 1997; Carletto 2022), particularly in low- and middle-income countries where much of refugee income is earned informally or through self-employment in, for instance, agriculture, services or trade. Despite this, income measures are the only ones that are available in some countries, most notably those in Latin America, where there is a tradition of using income approaches for measuring poverty. However, in most instances in this paper, a different approach is pursued, in which refugee earned income is approximated by deducting humanitarian assistance from refugee consumption. For the latter to proxy income, one needs to assume that refugee (net) savings and borrowing are negligible, while remittances are considered part of self-earned income. The measurement of income and consumption of refugees using surveys is an evolving field, and microdata sets are becoming increasingly available. For this paper, 11 microdata sets were brought together. To our knowledge, these 11 datasets present the universe of microdata available for low- and middle-income countries suited for estimating refugee poverty and income. Nine of these datasets also permit the estimation of poverty and income among host communities, with this data not available for Jordan and Bangladesh only. In a subset of economies (Ethiopia, Jordan, Kenya, the West Bank and Gaza, Uganda) the data can be broken down into refugees living in camps and those living elsewhere, which further enriches the analysis. 3 When income is invested in an asset that will last for multiple periods, only the use value of the asset is included in the consumption aggregate. 6 Income is estimated directly from the data sets for Bangladesh, Colombia and Costa Rica; it is proxied using consumption in all others. When deriving the income proxy the team paid special attention to how gifts are recorded. In certain surveys, this was in the consumption module itself; in others, it was collected through a special “assistance” module. Another area of attention was the value attached to shelter (a gift for many refugees and as such deducted from income). The value of essential needs is reflected by the global poverty lines for low-income countries (LICs), lower- middle-income countries (LMICs) and upper-middle-income countries UMICs. These are respectively PPP$ 2.15, PPP$ 3.65 and PPP$ 6.85. To convert 2017 PPP dollars into current US dollars, an exchange rate of 1.23 current US dollars for every PPP dollar is used (January 2024). Each country’s income classification is based on World Bank data as per the end of 2022. Table 1: Microdata sets used to estimate refugee income poverty gap Economy Survey Name Income Poverty Survey measured as line Year used Bangladesh (*) Cox’s Bazar Panel Survey Income 3.65 2023 Chad 4th National Harmonized Survey on Households’ Consumption 2.15 2018 Consumption and Informal Sector Colombia Gran Encuesta Integrada de Hogares (GEIH) Survey ; Income 6.85 2021 Encuesta Pulso de la Migracion Costa Rica Encuesta Nacional de Hogares (ENAHO) Income 6.85 2021 Ethiopia Socioeconomic Survey of Refugees in Ethiopia (SESRE) Consumption 2.15 2023 Jordan (*) Vulnerability Assessment Framework (VAF) survey for Consumption 6.85 2018 refugees Kenya Kalobeyi Socioeconomic Survey Consumption 3.65 2019 Kakuma Socioeconomic Survey Consumption 3.65 2021 Kenya Integrated Household Budget Survey Consumption 3.65 2016 Lebanon (**) Lebanon Vulnerability Assessment Panel (LVAP) Consumption 6.85 2022 Niger Enquête Harmonisée sur le Conditions de Vie des Consumption 2.15 2018 Ménages (EHCVM) West Bank and Consumption 6.85 Palestine Expenditure and Consumption Survey 2023 Gaza Uganda Refugee and Host Communities Household Survey Consumption 2.15 2018 (*) these surveys do not comprise data on host communities. (**) Lebanon became a lower-middle income country in 2022. Here the upper-middle income poverty line is preferred, mostly because the PPP conversion factor for Lebanon appears off leading to unrealistically optimistic self- reliance estimates. In addition, self-reliance estimates using the national poverty line (NPL) will be presented. 7 4. Global Results This section presents the results of the paper’s empirical analysis. By measuring poverty and self-reliance of refugees using the method outlined in section 3, the paper presents a number of key descriptives on these factors, with key statistics summarized in Table 2. 8 Table 2: Poverty and self-reliance among hosts and refugees (percent) Hosts Refugees Poverty Self- Poverty Own Own income Aid Self- incidence reliance incidence income poverty gap *** reliance IPL poverty** 2.15 Chad (*) 57.8 42.2 67.7 82.0 42.3 15.4 18.0 Ethiopia 24.9 75.1 74.8 85.5 65.0 32.2 14.5 Addis Ababa 17.8 82.2 7.3 9.1 2.1 0.8 90.9 in camps (*) 31.7 68.3 83.7 95.6 73.3 36.3 4.4 Niger 45.8 54.2 62.1 69.3 32.2 8.9 30.7 Uganda 37.5 62.5 69.5 85.9 56.3 28.0 14.1 Kampala 3.1 96.9 2.7 8.8 2.9 2.8 91.2 Outside Kampala (*) 41.3 58.7 74.2 91.3 60.1 29.8 8.7 3.65 Kenya 39.8 60.2 57.0 70.7 38.9 17.7 29.3 Kakuma (*) 87.1 12.9 69.9 83.7 49.5 22.9 16.3 Kalobeyei (*) 87.1 12.9 72.2 92.8 63.5 33.2 7.2 Dadaab (*) 61.0 39.0 44.5 61.4 27.9 12.1 38.6 Nairobi 22.5 77.5 33.1 33.9 9.8 0.5 66.1 Other urban 19.7 80.4 30.7 32.2 11.9 5.7 67.8 6.85 Bangladesh (*) NA NA NA 99.8 89.7 NA 0.2 Colombia 39.2 60.8 49.8 51.9 21.7 1.9 48.1 Costa Rica 13.4 86.6 23.6 27.2 10.1 2.0 72.8 Jordan NA NA 32.3 59.1 23.7 17.2 40.9 Amman NA NA 28.0 47.4 16.5 11.6 52.6 Outside Amman NA NA 41.8 61.3 26.2 16.9 38.7 Azraq (*) NA NA 19.5 77.5 34.6 32.2 22.5 Zaatari (*) NA NA 13.3 68.6 25.5 23.8 31.4 Lebanon (UMIC) (****) 2.3 97.7 16.4 30.2 7.2 4.3 69.8 Lebanon (NPL) 34.2 65.8 88.9 90.7 45.8 6.0 9.3 West Bank & Gaza 15.2 84.8 31.4 38.4 15.6 6.3 61.6 West Bank 5.1 94.9 5.7 8.4 1.7 0.8 91.6 Gaza 47.6 52.4 46.2 55.6 23.5 9.3 44.4 Note: (*) Indicates a refugee camp or settlement; (**) Refers to pre-assistance income poverty (***) Aid is expressed in percent of the poverty line (****) Survey data for Lebanon is available from the 2022 Vulnerability Assessment Panel. There are large variations in refugee poverty estimates for Lebanon ranging from 1 percent using the LMIC poverty line (PPP$ 3.65) and 16 percent using the UMIC poverty line (PPP$ 6.85) to 40 percent using the national poverty line. As Lebanon became a LMIC in July 2022, the first estimate should apply, even though, since Lebanon only recently became a LMIC a case could be made for using the UMIC poverty line. 4 4 Yet even the poverty estimate derived for the UMIC line is on the low side, certainly in comparison to other countries in the region. Lebanon’s low PPP conversion factor, which leads to low poverty lines, is behind this. For instance, the UMIC poverty line is only 38 percent of the national poverty line, whereas it is about 70 percent of the 9 Poverty among refugees is high, especially in LICs, with poverty incidence exceeding 60 percent in all countries examined, and reaching as high as 75 percent in Ethiopia. Poverty incidence among refugees is lower in MICs, varying from 50 percent in Kenya (a lower middle-income country) to 24 percent in Costa Rica (an upper-middle income country). Among the middle-income countries, Bangladesh takes an exceptional position. Though consumption poverty could not be determined, based on the estimates for income poverty and the income poverty gap, poverty among refugees is likely to be extremely high. Poverty among hosts – typically those living within 15 km of where refugees reside – is often also very high, and in some cases exceeds the poverty rates of local refugees. For instance, in Kakuma, Kalobeyei and Dadaab, poverty among refugees is lower than poverty among their hosts. Figure 1: Poverty incidence among hosts and refugees 100 90 80 75 68 70 70 62 58 57 60 50 Percent 50 46 38 40 39 40 32 31 30 25 24 20 13 15 10 0 Chad Ethiopia Niger Uganda Kenya Colombia Costa Rica Jordan West Bank & Gaza 2.15 3.65 6.85 IPL Hosts Refugees Aid received by refugees varies between countries. Table 2 presents aid received by refugees as a fraction of the poverty line. The data presented captures aid delivered to poor households only, with aid received by non-poor refugees (leakage) omitted from the analysis. Expressed as a percentage of the poverty line, aid received by poor households varies considerably. Poor refugees in Ethiopia receive 32 percent of the poverty line in assistance; those in Niger only 9 percent. On average, poor refugees in LICs receive more aid, but these averages mask significant within group variations. For instance, in Kenya and Jordan, refugees receive 17-18 percent of the poverty line, in Costa Rica and Colombia, only 2 percent. national poverty line in Jordan. Given this large uncertainty, data from Lebanon is excluded from further analysis in this section. 10 Figure 2 presents aid received by refugees in current US dollars. 5 Presented this way, stark differences in how refugees are assisted become visible. Poor refugees in Jordan receive US$ 529 per year, six to ten times more than those in Niger (US$ 86), Colombia (US$ 58) and Costa Rica (US$ 60). Figure 2: Aid received by poor refugees (expressed in US$) 600 529 500 400 311 290 270 US$ 300 194 200 149 86 100 58 62 0 Chad Ethiopia Niger Uganda Kenya Colombia Costa Rica Jordan West Bank & Gaza 2.15 3.65 6.85 IPL Aid reduces poverty among refugees significantly. As may be expected, the provision of aid reduces poverty among refugees considerably (see Table 2). The contribution to poverty reduction is largest in Jordan, where refugee poverty reduces by almost half, from 59 to 32 percent, thanks to the provision of aid. Nowhere else is the contribution of aid this significant. In Niger, where aid levels to refugees are low, aid leads to a decline in poverty incidence of just 7 percentage points; while in Costa Rica and Colombia, the effect is even smaller at 2-3 percentage points. Thus, there is a clear relation between the amount of aid spent and the reduction in poverty. Figure 3 uses this data to show how much it costs, on average per refugee, to reduce refugee poverty by one percentage point. These numbers should be interpreted with caution. Even when aid does not reduce poverty it may be spent well by alleviating the suffering of extremely poor refugees while being insufficient to bring them across the poverty line. With this caveat in mind, Figure 3 demonstrates the large variation that exists in the efficiency with which aid dollars can reduce refugee poverty. In Chad and Niger, US$ 10- 12 “buys” a percentage point reduction in refugee poverty. In Ethiopia, Colombia, and the West-Bank and Gaza, three times more money is needed for poverty incidence to fall by a percentage point. Interpreted differently, if poverty reduction was the only objective of giving aid to refugees, then a reallocation of aid resources between these economies would likely achieve far greater results. 5 All calculations convert these 2017 PPP dollars in current US dollars using an exchange rate of 1.23 current US dollars for every PPP dollar (January 2024). 11 Figure 3: Cost of reducing refugee poverty by 1 percentage point 35.0 29 28 30.0 28 25.0 21 20 20.0 16 17 US$ 15.0 12 10 10.0 5.0 0.0 Chad Ethiopia Niger Uganda Kenya Colombia Costa Rica Jordan West Bank & Gaza 2.15 3.65 6.85 IPL Refugees are less self-reliant than their hosts. Hosts are more self-reliant than refugees in all the economies examined in this paper (comparable data was not available for Bangladesh and Jordan). This should not come as a surprise. Refugees have typically lost many or all of their assets, are traumatized by their experiences, and frequently face significant labor market restrictions. However, the fact that as few as 14-15 percent of refugees in Uganda and Ethiopia, respectively, earn an income that takes them across the poverty line is concerning, even more so when 75 percent and 63 percent of host communities in Ethiopia and Uganda, respectively, are able to do so. It suggests that much of the productive capacity refugees embody is not utilized. This is in significant contrast to the UMICs of Costa Rica and Colombia, where the difference in self-reliance between hosts and refugees is much smaller. Figure 4: Self-reliance among hosts and refugees 100 98 87 85 90 75 73 80 70 66 70 63 60 61 62 60 54 Percent 48 50 42 41 40 31 29 30 18 20 15 14 9 10 0.2 0 Ethiopia Costa Rica West Bank & Gaza Kenya Bangladesh Colombia Jordan Uganda Lebanon (NPL) Niger Chad Lebanon (UMIC) 2.15 3.65 6.85 IPL Hosts Refugees Note: No data available for hosts in Bangladesh and Jordan 12 Large differences in self-reliance exist within countries. For Ethiopia, Uganda, Kenya and Jordan, data exist on refugees residing in camps and among host communities (Figure 5). This data allows one to consider self-reliance in slightly more detail. Two aspects stand out. Refugees residing in camps tend to have low levels of self-reliance. In urban areas, by contrast, the degree of self-reliance tends to be much higher. Self-reliance is highest in Addis Ababa and Kampala, while in Nairobi, where refugee (work) freedoms are restricted by an encampment policy and significant constraints on the right to work (Betts, 2021), self-reliance is noticeably lower. Under such conditions, it is not surprising that refugees in Nairobi are less self-reliant than those in the comparatively freer city of Kampala. Nor is it surprising that refugees in Addis Ababa have such high levels of self-reliance. Here, the country’s Out-of-Camp Policy requires refugees to demonstrate they can cover their cost of living and provide a sponsor before they are allowed to move out of camp, ensuring high levels of self-reliance (Kassa et al. 2019). Self-reliance is also higher among refugees in Jordan’s capital city, Amman, than in the country’s camp-based settings of Azraq and Zaatari. Yet refugee self-reliance in Jordan remains relatively low for a middle-income country. This is unsurprising given the restrictions the country imposes by requiring refugees to obtain work permits, which confine refugees to jobs in low-skill sectors only (Hoogeveen and Obi, 2024). In addition to clear disparities between urban and camp settings, significant variation exists between camps, with only 4 percent of refugees self-reliant in camps in Ethiopia, compared to 39 percent in Dadaab in Kenya. Such variation is also found within countries, with camps located in more isolated areas with fewer economic opportunities characterized by lower levels of self-reliance. For example, self-reliance in the isolated camp of Azraq is lower than in Zaatari which is 10km east of the northern Jordanian city of Mafraq. While in Kenya, self-reliance in Kakuma and Kalobeyei is less than in Dabaab, which is much closer to population centers and trade routes. Figure 5: Difference in refugee self-reliance by location within countries 100 91 91 80 68 66 60 53 39 41 39 40 29 31 23 15 14 16 20 9 4 7 0 All refugees All refugees All refugees All refugees Other urban Azraq (*) In camps (*) Dabaab (*) Nairobi Outside Amman Addis Ababa Outside Kampala (*) Kalobeiye (*) Kakuma (*) Zaatari (*) Amman Kampala Ethiopia Uganda Kenya Jordan Self-reliance is negatively correlated with aid. As shown in Figures 6 and 7, self-reliance is negatively correlated with aid at both national and sub-national levels. One way to interpret this is that aid provides a disincentive to work, an explanation made less plausible by the fact that the amount of aid is limited and 13 the degree of destitution high. What is more likely is that countries with severe restrictions on refugees’ ability to work and earn a living create aid-dependency. Indeed, as Figure 7 shows, comparisons within countries reveal that refugees in camps are less self-reliant and receive more aid compared to those in cities who are more self-reliant and receive less assistance. Figure 6: Aid and self-reliance of refugees 80 73 62 60 48 41 Percent 40 32 31 28 29 20 14 15 15 18 18 17 9 6 2 2 0 Uganda Ethiopia Chad Niger Kenya Jordan Colombia West Bank Costa Rica & Gaza 2.15 3.65 6.85 IPL Aid Self-reliance Figure 7: Aid and self-reliance by location within countries 100 40 80 30 Self-reliance 60 20 Aid 40 10 20 0 0 Kalobeiye Nairobi Outside Amman Outside Kampala Addis Ababa Kenya other urban Azraq Camps Zaatari Kakuma Amman Kampala Dabaab Ethiopia Uganda Kenya Jordan Aid Self-reliance Self-reliance is far more likely to reduce refugee poverty than aid. Figure 8 uses data from Table 2 to calculate how much it would cost to bring refugee income in line with the poverty line. The figure illustrates 14 three aspects. First, in all the economies, income earned by refugees exceeds aid received, challenging the common perception that refugees are passive recipients of aid. Second, despite significant amounts of aid being spend on refugees, in all the economies included in this study, aid is insufficient (on average) to bring refugee to or above the global poverty line. Finally, given a choice as how to best to bring refugees to the global poverty line, increasing refugee self-reliance holds more potential than increasing aid. In LICs, refugee incomes would need to increase by 62 percent on average to attain the poverty line. Aid, by contrast, would have to increase by 115 percent. In MICs, the difference is even greater, with refugee incomes needing to increase by 17 percent to reach the global poverty line; whereas aid would need to increase by 221 percent. This shows that aid, while essential in reducing the poverty levels experienced by refugees, is surpassed by the potential contribution refugee economic activity can have in increasing incomes and reducing poverty. Figure 8: Composition of refugee income needed to reach the poverty line 3500 3000 2500 2000 US$ 1500 1000 500 0 Chad Ethiopia Niger Uganda Kenya Colombia Costa Rica Jordan West Bank & Gaza IPL 2.15 3.65 6.85 Income Aid Shortfall 5. Conclusion This paper presents a straightforward approach to measuring refugee self-reliance, one that has several advantages over alternative measures that are available. It allows for refugee self-reliance to be determined through a quantifiable and widely understood metric: income generation, while anchoring refugee self-reliance within existing poverty measurements and SDG 1.1 (the elimination of extreme poverty). Theoretically, it offers an appropriate measure with which to conceptualize refugee needs, capturing both income and, at least partially, non-income factors in its measurement, while also appropriately accounting for ‘independence from aid’ by focusing on earned income only. As the analysis presented in this paper shows, refugee self-reliance is influenced by several factors. At the macro-level, location and economic environment are key. Refugees in middle-income countries are far more likely to be self-reliant than those in lower income countries, while refugees residing in urban and non-camp settings typically demonstrate far higher levels of self-reliance than those in rural and camp environments, reflecting the presence of greater economic opportunities outside camps and in urban 15 areas. Furthermore, while some labor market restrictions do not seem to have a significant impact on self- reliance, an inability to work significantly impairs refugee self-reliance. The paper also finds evidence for aid and self-reliance to be inversely correlated. In settings in which refugees are less likely to be self-reliant, levels of aid are higher; while in settings in which refugees are more self-reliant, levels of aid are lower. Whereas this correlation makes sense from a humanitarian perspective – aid needs to go where it is most needed, from a developmental perspective it suggests a perverse incentive, in that more aid is spent in environments where refugee self-reliance is hindered. Since aid and refugee self-earned incomes are substitutes, a different outcome can be envisaged in which governments that encourage refugees to be self-reliant receive more aid. Such an approach would allow for the reallocation of resources that are presently spent for humanitarian purposes for development purposes. This would not only be beneficial for refugees, who would gain financial autonomy and self- reliance, but would also be beneficial for hosts, who, as demonstrated in this paper, often face high levels of poverty and at times are even poorer than refugees. Investments in the development potential of hosting regions and their communities will thus benefit both refugees and hosts alike, increasing the incomes and self-reliance of both communities. 16 References Betts, A. (2021) The Wealth of Refugees. How displaced people can build can build economies. Oxford University Press. Betts, A.; Chaara, I.; Omata, M.; Sterck, O. (2020) The Kalobeyei Settlement: A Self-reliance Model for Refugees?, Journal of Refugee Studies, Volume 33, Issue 1, Pages 189–223, Betts, A.; Geervliet, R.; MacPherson, C.; Omata, N.; Rodgers, C.; Sterck, O. (2018) Self-Reliance in Kalobeyei? Socio-Economic Outcomes for refugees in North-West Kenya. July. Oxford RSC. Carletto G.; Marco Tiberti, M.;Alberto Zezza, A. (2022) Measure for Measure: Comparing Survey Based Estimates of Income and Consumption for Rural Households, The World Bank Research Observer, Volume 37, Issue 1, February 2022, Pages 1–38, https://doi.org/10.1093/wbro/lkab009 Deaton, A. (1997). The Analysis of Household Surveys: A Microeconometric Approach to Development Policy Baltimore, Maryland: Johns Hopkins University Press. Easton-Calabria, E. (2022) Refugees, Self-Reliance, Development: A Critical History. Bristol University Press. Easton-Calabria, E.; Omata, N. (2018) Panacea for the refugee crisis? Rethinking the promotion of ‘self- reliance’ for refugees, Third World Quarterly, 39:8, 1458-1474, DOI: 10.1080/01436597.2018.1458301 Foster, J.; Joel Greer, J.; Erik Thorbecke, E. (1984) A class of decomposable poverty measures, Econometrica. 52(3): 761–766. doi:10.2307/1913475. Greeley, M. (1994) Measurement of Poverty and Poverty of Measurement, IDS Jolliffe, D. M.; Gerszon Mahler, D.; Lakner, C.; Atamanov, A.; Tetteh Baah, S. K. (2022) Assessing the Impact of the 2017 PPPs on the International Poverty Line and Global Poverty. The World Bank. Tadesse, K.; Mulatu, F.; Iddossa, J. (2019) Ethiopia's Refugee Policy Overhaul: Implications on the out of Camp Policy Regime and Rights to Residence, Movement and Engagement in Gainful Employment. Ethiopian Journal of Human Rights. Vol 4: pp124-167. Ravallion, M.; Datt, G.; van de Walle, D. (1991) Quantifying Absolute Poverty in the Developing World. Review of Income and Wealth vol. 37(4):345–61. Schön, A. M. (2020) Assessing the Level of Self-Reliance and Livelihood of Encamped Refugees. Syrian Refugees in Jordan Labor and Globalization, edited by Christoph Scherrer, Volume 20, ISBN 978-3-95710- 281-2, Rainer Hampp Verlag, Augsburg, München 2020 Sen, A. (1999). Development as freedom. New York, NY: Anchor Books SRI (n.d.) Refugee Self-Reliance Initiative homepage, Available at: https://www.refugeeselfreliance.org/sri [Accessed 24.05.2024] UNHCR (2005) Handbook for Self-Reliance. Geneva: UNHCR. UNHCR (2014) A Global Strategy for Livelihoods: A UNHCR strategy 2014-2018 UNHCR (2023) Global survey on livelihoods and economic inclusion report (December 2023). 17 UNHCR (2024a). Refugee Data Finder: https://www.unhcr.org/refugee-statistics/ accessed May 30, 2024. UNHCR (2024b). Sustainable Programming. What is Sustainable Programming And Why Is It Needed? https://reporting.unhcr.org/libraries/pdf.js/web/viewer.html?file=https%3A%2F%2Freporting.unhcr.org% 2Fsites%2Fdefault%2Ffiles%2F2024-06%2FSustainable%2520programming%2520paper%2520- %252017%2520June%25202024.pdf. Accessed August 20, 2024. 18