Displaced During Crisis Lessons Learned from High-Frequency Phone Surveys and How to Protect the Most Vulnerable A © [2023] International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy, completeness, or currency of the data included in this work and does not assume responsibility for any errors, omissions, or discrepancies in the information, or liability with respect to the use of or failure to use the information, methods, processes, or conclusions set forth. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be construed or considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. Images: © UNHCR. Used with the permission of UNHCR. Further permission required for reuse. Cover design: Florencia Micheltorena Table of Contents Acknowledgement iv List of Acronyms v Executive Summary vi 1. Introduction 1 2.  Microdata and Methodology 5 2.1  High-Frequency Phone Surveys during COVID-19 6 2.2  Phone Survey Sampling Strategies among Displaced Populations 9 2.3 Biases 11 2.4  Harmonization of the HFPS Data 12 3.  The Welfare of Displaced and Host Populations during the Pandemic 17 3.1  Pandemic Context 18 3.2 Employment 18 3.3 Income 25 3.4  Assistance and Coping 27 3.5  Food Security 29 3.6  Education and Learning 35 4.  Financing For Displaced Populations 42 5.  Discussion and Conclusion 48 Annexes Annex 1. Country Surveys 54 Annex 2. Supplemental Figures and Tables 68 Annex 3. Regression Tables 79 Annex 4. Estimating Aid for Displaced Populations Using OECD CRS Data 82 References 89 i Figures Figure 1.1  Stock of Forcibly Displaced Populations 2000–22 2 Figure 1.2  Share of FDPs by host country income group, 2019 3 Figure 3.1  Share of Employed by Host and FDP Type, before and during the Pandemic (%) 19 Figure 3.2  Marginal effect of FDP type on probability of work during the pandemic 21 Figure 3.3  Marginal effect on probability of stopping work 21 Figure 3.4  Marginal Effect on Probability of Working during the Pandemic for Refugees 22 Figure 3.5  Employment Transitions across Employment Types, by Host and FDPs 24 Figure 3.6  Employment Transitions across Sector of Activity, by Host and FDPs 24 Figure 3.7  Share of Households with Work Stoppages and Income Losses, by Population Group (%) 26 Figure 3.8  Work Stoppages and Income Losses, by Camp Status 26 Figure 3.9  Share of Households Reporting Income Losses, by Source of Income (%) 26 Figure 3.10  Marginal Effect of Being a Refugee on the Probability of Income Loss during the Pandemic 27 Figure 3.11  Share of Households Receiving Assistance during the Pandemic, by Host and FDPs (%) 28 Figure 3.12  Share of Households that Relied on Various Coping Measures, by Host and FDPs (%) 29 Figure 3.13  Agricultural Price Indices 30 Figure 3.14  Number of People Facing Crisis-Levels of Acute Food Insecurity 30 Figure 3.15  Estimates from the Linear Probability Model on the Likelihood of Running out of Food 31 Figure 3.16  Share of Households with Adults Having Skipped a Meal Because of Lack of Resources (%) 32 Figure 3.17  Share of Households with Members Not Having Eaten for a Day Because of Lack of Resources (%) 32 Figure 3.18  Marginal Probability of Displaced Households with Members Not Having Eaten for a Day Because of Lack of Resources, Relative to Host Households 33 Figure 3.19  Food Insecurity Experience Scores 33 Figure 3.20  Length of Full or Partial School Closure, Average Number of Weeks by Region 36 Figure 3.21  Length of Full or Partial School Closure between February 2020 and March 2022 by Country 36 Figure 3.22  Distance Learning Modalities Adopted, by Country Income Group 36 Figure 3.23  Share of Households with Children Who Stopped Learning during the Pandemic, by Country (%) 38 Figure 3.24  Share of Households with Children Accessing Education before and during the Pandemic, by FDP Type (%) 39 Figure 3.25  Prepandemic Schooling vs Stopped Learning during the Pandemic, by Country (%) 40 Figure 3.26  Schooling before the Pandemic vs Learning during the Pandemic among Refugees, by Country (%) 41 Figure 4.1  The Fiscal Response during COVID-19, by Country and Country Groups (% GDP) 44 Figure 4.2  Net ODA Received in 2021 (%) 44 Figure 4.3  Trend in Aid for Displaced Situations in Recent Years (constant 2020 US$ million) 45 Figure 4.4  Trend in Total Aid Flows in Recent Years (constant 2020 US$ million) 46 Figure 4.5  Aid to displaced populations, as proportion of total aid, by region (%) 46 Figure 4.6. Aid to displaced populations, per displaced person, by region (constant 2020 US$) 47 Figure 5.1  Poverty Trends by Country Groups, 2015–30 (%) 50 Figure A2.1a  Thirty LMICs Hosting the Most FDPs in 2019 69 Figure A2.1b  Thirty LMICs Hosting the Most FDPs as a Share of National Population in 2019 69 Figure A2.2  Mobility Trends and Policy Stringency in Countries with Phone Surveys 71 Figure A2.3  Share of Employed by Host and FDP Type, by Country (%) 72 Figure A2.4  Share of Households with Respondent Who Stopped Working during the Pandemic, by Host and FDP Type, by Country (%) 73 ii Figure A2.5  Share of Households that Received Any Social Assistance Since Pandemic Started, by Host and FDP Type, by Country (%) 74 Figure A2.6  Share of Households Receiving Assistance during the Pandemic, by Camp Status (%) 75 Figure A2.7  Share of Households that Ran out of Food Because of a Lack of Money or Other Resources in the Past 30 Days (%) 76 Figure A2.8  Share of Households with Household Members Not Eating for a Day due to Lack of Resources (%) 77 Figure A2.9  Share of Households with Children Accessing Education before and during the Pandemic, by Country and FDP Type 78 Figure A2.10  Share of Households with Children Accessing Education before and during the Pandemic, by Country and Camp Status (%) 79 Figure A4.1  Aid to Displaced Populations, Globally and by Country Groupings (in 2020 million US$) 87 Figure A4.2. Total Aid by country (in 2020 million US$) 88 Figure A4.3. Aid per displaced person by country (in 2020 US$) 89 Tables Table 2.1  Accommodation in Camps 8 Table 2.2  Available Samples in the Harmonized Data 8 Table 2.3  HFPS Design for Displaced and Host Samples 10 Table 2.4  Timing of Phone Surveys 14 Table 2.5  Descriptive Statistics for Harmonized Survey Data Respondents 15 Table 3.1  Employment Type Transition Matrix, by Hosts and FDPs 22 Table 3.2  Sector Transition Matrix, by Hosts and FDPs 23 Table 3.3  Food Insecurity Components 34 Table A2.1  Core Modules for the COVID-19 HFPS 70 Table A3.1a  Probability of Working 80 Table A3.1b  Probability of Income Loss 81 Table A3.1c  Probability of Not Eating for a Day 81 Table A3.2  Country-pooled Linear Probability Models on Select Outcomes 82 Table A4.1  Coverage of OECD CRS Database, Select Countries, 2019 84 Table A4.2  Keywords Used to Estimate Aid for Displaced Populations and Examples 84 Table A4.3  Detailed Sector Codes in the OECD CRS Emergency Response Sector 85 Table A4.4  Top ten recipient countries in terms of aid for displaced populations, 2016-2021 (in 2020 million $US) 86 Boxes Box 2.1  Use of Multiple Sampling Frames in Kenya and Bangladesh 9 iii Acknowledgement This report was co-led by Yeon Soo Kim and Jeffery Tanner, both Senior Economists in the global unit of the Poverty and Equity Global Practice. The core team comprises Harriet Mugera (Senior Data Scientist, DEC Analytics and Tools), Saad Imtiaz (Consultant), and Gildas Deudibe (Consultant). Petra Kaps (Consultant, JDC) provided additional research support. The preparation of the harmonized database used in this report would not have been possible without the diligent work and support provided by the Poverty and Equity Global Practice’s Data for Goals team, in particular, Ifeanyi Edochie (Data Scientist, Poverty and Equity Global Practice), who coordinated the harmonization work. The country teams that originally collected the phone survey data have also provided invaluable support. The work was carried out under the overall direction of Luis Felipe Lopez Calva (Global Director, Poverty and Equity Global Practice) and Benu Bidani (Practice Manager, Poverty and Equity Global Practice). The team gratefully acknowledges feedback from Utz Pape (Senior Economist, Poverty and Equity Global Practice), Sharad Tandon (Senior Economist, Poverty and Equity Global Practice), and Alia Al-Khatar-Williams (Deputy Director, UNHCR), who served as peer reviewers for the report, as well as Thomas Ginn (Research Fellow, Center for Global Development). iv List of Acronyms CBPS Cox’s Bazar Panel Survey (Bangladesh) CONASUR Conseil National de Secours d’Urgence et de Réhabilitation (National Council for Emergency Relief and Rehabilitation) CRS Creditor Reporting System (OECD) DAC Development Assistance Committee (OECD) DRC Democratic Republic of Congo DWRAP Developing World Refugee and Asylum Policy FDP Forcibly Displaced Population FIES Food Insecurity Experience Score GCFF Global Concessional Financing Facility GCR Global Compact on Refugees HFPS High-Frequency Phone Survey IDA International Development Assistance IDP Internally Displaced Person IOM International Organization for Migration JDC Joint Data Center (World Bank-UNHCR) LIC Low-Income Country LMIC Low- or Middle-Income Country MIC Middle-Income Country NGO Nongovernmental Organization NPM Needs and Population Monitoring (IOM) ODA Official Development Assistance OECD Organization for Economic Cooperation and Development OLS Ordinary Least Squares PoC Person of Concern ProGres Profile Global Registration System (UNHCR) RDD Random Digit Dialing RRPS Rapid Response Phone Survey RSW Regional Sub-Window SES Socioeconomic Survey UNHCR United Nations High Commissioner for Refugees VDA Venezuelans Displaced Abroad WFP World Food Programme v Executive Summary A woman walks past a puddle created vi by recent rain in Dori, Burkina Faso. © UNHCR/Nana Kofi Acquah, June 2021  T he world is emerging from a series of shocks that led to widespread turmoil in lives and livelihoods. The COVID-19 pandemic generated the worst economic downturn since the Second World War and The newly harmonized database represents a rich source of information in a context where there has been little coordinated research on how systemic shocks differentially affect forcibly displaced and host populations. had a disproportionate impact on the poor and vulnerable. Following the initial shock, the recovery The evidence from this new database shows that was similarly uneven and was further hampered by FDPs were deeply affected by the pandemic and a cost-of-living crisis that quickly unfolded as food that they often, though not always, fared worse and energy prices skyrocketed. than their hosts. FDPs typically experienced larger initial employment losses that were then Although historically low global poverty figures followed by a slower recovery. In addition, there before the pandemic reflect a steady decline were significant job changes among those who over several decades, extreme poverty has been remained employed, again with greater turnover increasingly concentrated in Sub-Saharan Africa among FDPs. Household income dynamics, where and in fragile and conflict-affected countries. available, suggest that the welfare impact was The latter set of countries host about 10 percent much more widespread than indicated by outright of the global population but nearly 40 percent employment losses alone. Although labor income of the global poor. Understanding the welfare of losses were most common, in some countries, a high vulnerable populations, including during times of share of FDPs reported reductions in assistance, economic shocks, is therefore critical to addressing an important source of income. Food insecurity— threats to the trajectory of global poverty and not a new challenge for many countries that host shared prosperity. displaced populations—reached alarming levels during the pandemic, with FDPs almost always Amid the devastating impacts of the pandemic, reporting worse outcomes. Efforts to support those the crisis created an opportunity for a large- in need likely fell short, leaving much of the negative scale data collection effort on forcibly displaced welfare shock unmitigated. On top of the economic populations (FDPs)—a group on which there setbacks, hard-earned gains in education were lost exist significant data gaps. This started out as a during long school closures. series of country-level efforts that served as the basis for a newly harmonized database of phone In many ways, the pandemic exacerbated an surveys from 14 countries during the first two years already precarious situation at the same time that of the COVID-19 pandemic. This contemporaneous other preexisting and contemporaneous factors database of host and displaced populations offers were contributing to a deteriorating welfare unique insights into the welfare of FDPs relative trend among FDPs. The pandemic worsened the to their hosts, while also allowing for comparisons welfare of displaced populations who are already between different populations of concern among the poorest and most vulnerable groups. (internally displaced persons, refugees, hosts) and Refugees often do not have full legal rights to accommodation types (in camps, out of camps). work in their host countries, and the absence of vii such rights, unsurprisingly, is correlated with lower reliance on humanitarian assistance. Greater employment levels across the board. In addition economic opportunities for FDPs will reduce the to pandemic-related disruptions, there were often burden of hosting, and granting them formal labor other contemporaneous factors that adversely market access can be a positive first step—indeed, affected both displaced and host populations. This labor market participation tends to be higher was particularly the case with food security–related in countries that allow work rights for refugees. outcomes, where in addition to rapidly rising Another key means of integration and the promotion global food prices, some countries also faced local of self-reliance is providing refugee children access preexisting or concurrent challenges, including the to national education systems and relieving the arrival of cyclical lean periods, fuel price shocks, many social and economic constraints to their and the escalation of conflict and violence. learning. Although remedial support is needed by all, displaced children are in a more disadvantaged The recent welfare losses raise concerns that position due to their lack of financial stability and the effects of the pandemic could mean higher heightened vulnerability. Displacement status can poverty and inequality for a generation—not only be an easy indicator for identifying one group of among FDPs but also their hosts. This would be particularly vulnerable children in need of targeted particularly the case if the losses are not alleviated catch-up learning. over time. Lost assets and savings take time to rebuild. Extensive learning losses during COVID Sustainable financing solutions that allow for could be compounded as pandemic-affected continued investments and longer-term planning generations enter the labor market and their future will be critical to easing the burden on major earnings are further depressed. hosting countries. Considering the record- high levels of displacement and its increasingly Inclusive policies and support for the self-reliance protracted nature, financing needs are not likely of displaced populations can shape this into a to diminish soon. Many host countries rely heavily very different trajectory. Displaced populations on official development assistance for government create significant social, economic, and political spending and for supporting displacement pressures on the host countries, which are situations in their countries. As learned the hard predominantly made up of low- and middle-income way during the pandemic, a key challenge of countries, many of which are struggling with their current displacement financing is that it may own development challenges, including high not be available when needed most. Financing debt and low growth. During the pandemic, host arrangements need to be predictable and reliable countries were often ill-equipped to extend support for planning purposes beyond the short term. to displaced populations as they were constrained The World Bank’s International Development by tightened fiscal space to respond to COVID’s Association (IDA) Window for Host Communities devastating impacts on the general population. and Refugees can help with FDPs crossing national To make matters worse, external aid for FDPs borders, but similar financial support does not exist declined in 2020 during the most acute phase for the far more numerous internally displaced of the pandemic, even as overall aid increased. populations. Shifting the balance of support more Because repatriation is rare, commitment to the toward development aid and adopting more burden sharing outlined in the Global Compact for inclusive policies for the displaced can help ease Refugees is critically needed. Similarly, external the overall burden of hosting. support can help countries working toward durable solutions for internally displaced populations. Finally, the complex nature of the challenges presented by displacement situations Supporting policies that will aid FDPs in becoming underscores recent calls for statistical inclusion more self-reliant will help build their productive to provide more and better data that can be capacity and resilience, which in turn will reduce relied upon to design better policies. Despite the financial burden on host countries and their technical and budgetary challenges, including viii FDPs in data collection efforts is often best done on Refugee, IDP and Statelessness Statistics in collaboration with national statistical offices. (EGRISS). Because data collection on FDPs requires These results demonstrate the value of open, reliable sampling frames, up-to-date and complete harmonized, longitudinal data on displaced registration databases are invaluable. Formal data populations to monitor periods of crisis and sharing agreements can facilitate institutional recovery. Harmonization would be greatly aided by exchanges. The phone survey experience during using standardized survey instruments, particularly the COVID-19 pandemic shows that it is feasible to as they integrate the UN Statistical Commission’s collect data during crisis that is not only statistically recommendations developed by the Expert Group robust but also time and cost efficient. ix CHAPTER 1 Introduction A man towards the mass grave where his son is buried x in Plain Savo site, Democratic Republic of the Congo. © UNHCR/Hélène Caux, March 2022  T he world is still emerging from compounding crises that have led to widespread turmoil in lives and livelihoods. The COVID-19 pandemic generated the worst global economic downturn, the largest total stock of FDPs at roughly 79 million. This included 26 million refugees, 46 million internally displaced persons (IDPs), 4 million asylum seekers, and 4 million other persons in need of international protection, including Venezuelans Displaced setback in the fight against global poverty, and Abroad (VDAs). After remaining stable at around possibly the largest single-year increase in global 40 million, the stock of FDPs exhibited a steep inequality since the Second World War, as income increase beginning in 2013, following a series of losses of the world’s poorest were twice as high displacement spikes in Afghanistan, the Levant, as those of the world’s richest (World Bank 2020c). Myanmar, the Sahel, and Venezuela. The major The number of extreme poor rose by over 70 million source of FDP growth over the past decade has in 2020 alone, increasing the global total to over been from IDPs, whose number rose sevenfold 700 million. Economic activities gradually resumed between 2005 and 2019. More recently, the war in around the world after the initial shock subsided Ukraine added over 8 million refugees to the total and lockdowns were lifted, but the recovery was (Figure 1.1).1 hampered by a cost-of-living crisis that quickly unfolded as food and energy prices skyrocketed. Displacements rose further through 2022, Three years after the onset of the COVID-19 exceeding a record-breaking 100 million, though pandemic, the recovery is still incomplete and has largely due to factors unrelated to the pandemic. been very uneven across countries and population COVID had a temporary impact on the displacement subgroups (World Bank 2022). trend, as pandemic-induced movement restrictions and border closures are estimated to have led Understanding the welfare of vulnerable to approximately 1.5 million fewer refugees and populations during times of economic shock is asylum seekers in 2020 than would have been critical to addressing threats to shared prosperity. expected without COVID (UNHCR 2021b). In the Forcibly displaced populations (FDPs) are some of same year, the number of asylum applications the poorest and most marginalized people in the fell by about a third in Organization for Economic world. Although historically low global poverty Cooperation and Development (OECD) countries figures before the pandemic reflect a steady but rebounded quickly in 2021 (OECD 2021, decline over several decades, extreme poverty has 2022). The latest trend suggests that the growth been increasingly concentrated in Sub-Saharan in refugee populations during the pandemic was Africa and in fragile and conflict-affected countries. largely a continuation of a preexisting secular The latter set of countries host about 10 percent of trend; for example, the crises that prompted large- the global population but nearly 40 percent of the scale displacement in Burkina Faso and Ethiopia global poor (World Bank 2022). were largely driven by causes not connected to COVID. As before the pandemic, fragility, conflict, Just before the pandemic hit, the United Nations and violence remain the leading causes of forced High Commission for Refugees (UNHCR) put the displacement (Corral et al. 2020).2 1  See UNHCR, “Refugee Data Finder,” https://www.unhcr.org/refugee-statistics/. 2 This intersects with displacement due to disasters and adverse effects of climate change (UNHCR 2021b); indeed, before the pandemic there were 5.1 million internally displaced persons as a result of environmental disasters in the world (IDMC 2020). 1 Figure Figure1.1 Stock 1 of Forcibly Displaced Populations 2000–22 120 100 80 Millions 60 40 20 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 Refugees (including Palestinian Refugees) Asylum Seekers IDPs (IDMC)** Others INIP including VDA Note: “Others INIP” denotes others in need of international protection (INIP), comprised mostly of Venezuelans Displaced Abroad (VDA). Source: Authors’ calculation from UNHCR Data Finder for refugees, asylum seekers and Others INIP; and the Internal Displacement Monitoring Centre (IDMC) for IDPs. UNHCR collects IDP data only for individuals who receive assistance and/or protection from the organization, whereas the IDMC offers a broader overview of internal displacement on a global scale. See UNHCR, “Refugee Data Finder,” https://www.unhcr.org/refugee-statistics/; and IDMC (2020, 2021, 2022). Protracted displacement situations have also Many of these major hosting countries were become increasingly common. The number struggling with low growth, high debt, and other of protracted refugee situations—defined as development challenges, making them ill-prepared populations that have been displaced abroad to extend support to displaced populations as for more than five consecutive years—has been they were buffeted by compounding global stable, following little growth during the pandemic. shocks. Growth in some countries was slowing or About 15.9 million people, or 74 percent of the regressing even before the pandemic: GDP per global refugee population, found themselves in capita increased in real terms between 2010 and long-lasting situations by the end of 2021 (UNHCR 2019 in only about a quarter of the 30 largest hosting 2022f). countries. In Chad, for example, GDP per capita in 2010 declined from US$728 in 2010 to US$653 in Displaced populations are concentrated in 2019 (constant 2015 US$). GDP growth cratered countries with low levels of development. As in 2020 in these and other countries, in line with illustrated in Figure 1.2, by the end of 2019, 43 global trends; for example, GDP contracted by 9 percent of FDPs were hosted in low-income percent in Iraq and by around 8 percent in Ecuador countries (LICs), and about half were in middle- and Mexico. Yet the growth rate in Ethiopia, for income countries (MICs). Similarly, more than three example, fell only slightly from 8.4 percent in 2019 in five IDPs were in LICs, and very few were in high- to 6.1 percent in 2020.4 Government debt stock as income countries. About 82 percent of refugees a share of GDP rose significantly between 2010 and (including asylum seekers and VDAs) lived in low- 2019, rising by over 50 percent in countries, such as or middle-income countries (LMICs). As of 2019, Uganda, Kenya, Djibouti, Burkina Faso, and Chad. 10 countries accounted for just 0.7 percent of This was broadly in line with regional and global global GDP, but they hosted one-third of the global trends and was followed by a global debt increase displaced population.3 Nearly two in five FDPs are of around 8.6 percentage points in 2020.5 The found in Sub-Saharan Africa (UNHCR 2019a), the fiscal balance for developing economies similarly region with the highest poverty rate (World Bank deteriorated, declining by an average of 2.8 percent 2022). per year in the 2010–19 period and dropping further 3 UNHCR, “Refugee Data Finder,” https://www.unhcr.org/refugee-statistics/; World Bank, “Macro Poverty Outlook,” https://www. worldbank.org/en/publication/macro-poverty-outlook. 4 World Bank, “World Development Indicators,” https://databank.worldbank.org/source/world-development-indicators. 5 World Bank, “Macro Poverty Outlook,” https://www.worldbank.org/en/publication/macro-poverty-outlook. 2 by 8.1 percent in 2020. This strongly implies that Figure 1.2 Share of FDPs by host country income countries had little buffer against negative shocks group, 2019 heading into the pandemic. Share of all FDPs in host countries by income classification, in 2019 The devastating impacts of the pandemic precipitated the need for wide-spread collection 8% of socioeconomic data on the displaced. Significant data gaps on this population remain, 43% 29% but taken together, these data collection efforts were unprecedented in their scale and so form the basis of this report. This endeavor started out as a series of country-level efforts, a large 20% number of which were supported by the World Bank-UNHCR Joint Data Center (JDC) on Forced Displacement. The newly harmonized database Share of IDPs in host countries by income classification, in 2019 developed for this report consists of phone surveys 0.5% fielded in 14 countries during the roughly two-year period of the COVID-19 pandemic, from March 2020 through December 2021. The resulting 23% database of contemporaneous host and displaced populations offers unique insights into the welfare 61% of a large number of FDPs relative to nondisplaced 16% populations and complements other recent efforts to build representative, harmonized surveys from LMICs. For example, a series of briefs by the World Bank (2023a, 2023b, 2023c) presents findings Share of Refugees, Asylum-Seekers, and Venezuelans displaced from harmonized surveys fielded from 2015-20 and abroad in host countries by income classification, in 2019 covers representative samples of displaced and host populations in 10 countries. 18% The new database is a rich source of information in 27% a context where there has been little coordinated research on how systemic shocks differentially 18% affect forcibly displaced and host populations. Analysis using this harmonized data allows for 36% robust comparisons across countries that could help identify systemic challenges. Conversely, such data can also help illustrate where there is heterogeneity High income hosting 2019 in experiences and identify outliers that can be Lower-middle income host countries 2019 Upper-middle income host countries 2019 probed to understand important deviations from Low-income host countries 2019 observed trends. Because it affected most countries at roughly the same time and in broadly similar Source: Staff illustration using “Refugee Data Finder,” https:// ways, the global economic shock associated with www.unhcr.org/refugee-statistics/. COVID—including local restrictions on movement and global hikes in food and commodity prices— This report makes several contributions to the provides an opportunity to better understand how literature on the welfare of FDPs and their hosts. On FDPs and their hosts are affected by the complex the data front, the harmonized database compiled dynamics of systemic shocks. Previous analysis for this report represents a large and unique source of earlier rounds of unharmonized phone survey of information on the welfare of both hosts and FDPs data from eight countries had illustrated how the during the period of an unprecedented pandemic. socioeconomic well-being of many FDPs and host The data span 14 countries from different regions, populations deteriorated during the first year of the populations of concern (IDPs, refugees, hosts), and pandemic (Tanner et al. 2021). accommodation types (in camps, out of camps). 3 Over a fifth of the global population displaced The report also brings a policy lens to the analysis. before the start of the pandemic is represented By examining the role of existing labor market in this database, allowing for direct comparison and education policies in the hosting country and aggregation of results across countries and using newly available information from a cross- subgroups. country, up-to-date policy database, the data yield important insights. In addition, the report examines Compared to previous studies, the analysis aid financing trends using disbursement-level data has been deepened to provide a more holistic from OECD’s Creditor Reporting System (CRS) and view of welfare among the displaced during the a labor-intensive keyword search approach to tease pandemic. The results from most country-level out disbursements that are intended for displaced High-Frequency Phone Surveys (HFPS) were populations. often intended to provide a quick snapshot of the pandemic’s welfare impact in a single country. This The rest of the report is organized as follows. report extends that existing analysis by providing Section 2 describes the data used in this exercise, a more comprehensive view of how welfare including the samples, the harmonization process, evolved during COVID for both hosts and displaced and the resulting database. Section 3 presents the populations in countries across the globe, and key results on the welfare impact of the forcibly thus contributes to a long literature on the general displaced and their hosts during the pandemic and welfare impacts of the pandemic (see Brunckhorst, ensuing crises. Where relevant and possible, the Cojocaru, and Kim, forthcoming, for a summary). It results are linked to preexisting sectoral polices also serves as a complement to a recently published that were in place in hosting countries before the World Bank report that analyzes global welfare pandemic. Section 4 discusses the recent trend in during the COVID-19 pandemic (Brunckhorst et official development assistance (ODA) intended for al. 2023) by lending a displacement lens to the displaced populations. The report concludes with a analysis. The labor market analysis, for example, discussion of the results and a set of forward-looking has been deepened to consider outcomes beyond policy recommendations, focusing on inclusive job losses, such as the extent of job changes social policies and sustainable financing aimed (similar to Brunckhorst et al. 2023 for nondisplaced at promoting self-reliance among the displaced, populations). and on lessons learned for data collection and harmonization following this unprecedented endeavor. 4 Microdata and CHAPTER 2 Methodology Four-year-old poses with her parents in Plain Savo, Democratic Republic of Congo. © UNHCR/Hélène Caux, March 2022 5 2.1  High-Frequency Phone Surveys During the COVID-19 crisis, HFPS approaches helped overcome the challenges of face-to-face during COVID-19 surveys that were suspended by restrictions enforced to mitigate the spread of the virus. HFPS Data allowing for comparisons between host had a nascent but strong track record of being and displaced populations are scarce. Displaced deployed in remote or risky areas, and with the people living outside of camps are rarely identified rapid acceleration of digital adoption in developing in questionnaires or picked up in sufficient numbers countries, technological innovations in survey to generate reliable subgroup statistics. Camps for administration have emerged in the past decade displaced populations are frequently overlooked in that have facilitated these alternative techniques. sampling strategies for national household surveys. In particular, “with the availability of inexpensive Although humanitarian and development agencies phone handsets and rapidly growing network have made great strides, the statistical inclusion coverage in many developing countries, the mobile agenda necessitates identifying vulnerabilities and phone has attracted much of the attention as a new addressing the humanitarian and development tool for collecting high-frequency and, oftentimes, challenges of marginalized communities. Including low-cost survey data” (Dabalen et al. 2016). displaced populations in data collection is essential to accurately understanding welfare; not doing A series of HFPS initiatives were implemented so may lead to significant underestimation of beginning in April 2020 in several developing population needs.6 Even so, accurately estimating countries. These were mainly led by the World welfare for displaced populations comes with Bank, in collaboration with other key stakeholders, significant methodological challenges. and aimed to capture the socioeconomic welfare of national populations during the pandemic. In some The extent and length of the pandemic increased of these countries, the surveys were extended need for regular and timely data to assess the to include FDPs. These FDP-related efforts were socioeconomic impacts of COVID-19, especially mainly led by the World Bank and UNHCR, often on vulnerable groups such as FDPs, who remain with support of the JDC. largely unaccounted for in household surveys. Indeed, despite the considerable number of The team applied a series of selection criteria forcefully displaced worldwide, the collection of when considering datasets to include in this good quality microdata on this particular population report. Eligible datasets included phone surveys7 remained limited (Dang and Verme 2022). This that were carried out by the World Bank or others prompted a call for action from the World Bank- during the pandemic from April 2020 through UNHCR JDC in its working paper opportunely December 2021 and that included representative,8 entitled, Highly Vulnerable Yet Largely Invisible: contemporaneous9 data on host10 and displaced Forcibly Displaced in the COVID-19-Induced populations (refugees11 or IDPs) in LMICs. To Recession (Vishwanath, Alik-Lagrange, and mitigate sampling and selection biases, the data Aghabarari 2020). 6  For example, current global poverty counts are often based on the assumption that the distribution of welfare among displaced people living in camps is equal to that of the rest of the country in which they reside in (Corral et al. 2020). Because FDPs are likely to have higher poverty rates, this assumption may underestimate poverty (Beegle and Christiaensen 2019). Studies conducted in Iraq, Peru, Somalia, South Sudan, and Uganda suggest that FDPs have roughly 25 percent lower welfare than the nondisplaced host population in the country (Sharma and Wai-Poi 2019; Pape and Parisotto 2019; Pape and Wollburg 2019; World Bank 2019b). Corral et al. (2020) estimate that the tendency of displaced people to be poorer than nondisplaced populations could raise the global poverty count by 33 million people. 7  Note that Jordan fielded both phone and face-to-face interviews simultaneously in select rounds of data collection to facilitate testing for modality effects. 8  The surveys used a robust probability sampling approach with a clear sample frame for the relevant population. Sample sizes must be sufficiently large to allow for statistical testing of differences between host and displaced populations at reasonable levels of discriminant validity. 9  Surveys needed to have at least one wave of a comparable phone survey of the host population and FDPs fielded in the same or adjacent months. 10  Reflecting the lack of international agreement on the definition of the term, “Host population” is used somewhat loosely here to mean the nondisplaced national population or the sub-populations that live within the administrative region of the country as the displaced group, as in Cox’s Bazar, Bangladesh, for example. See Table 2.3 for details for each country. 11  When referring to the survey data or results derived therefrom, “refugees” refers collectively to the population of refugees, asylum seekers, and Venezuelans Displaced Abroad (VDA) appropriate to each country’s context and sampling strategy. 6 were required to include sampling weights, or Country Abbreviations sufficient documentation to reconstruct sampling weights to make the data as representative as BFA Burkina Faso possible of the general host and the displaced BGD Bangladesh populations. Finally, questionnaires were assessed CRI Costa Rica to see which were sufficiently close to the World Bank’s core HFPS questionnaire to allow the data DJI Djibouti to be harmonized. DRC Democratic Republic of Congo ECU Ecuador The resulting survey catalogue represents a large source of information on the welfare of both ETH Ethiopia host and displaced populations in developing IRQ Iraq countries during the pandemic. It harmonized data across different regions, populations of concern JOR Jordan (IDPs, refugees, hosts), and their accommodation KEN Kenya type (in camps, out of camps)12 where available. MEX Mexico These data still had shortcomings – often through incomplete frames wherein not all geographic SOM Somalia areas were sampled, for example. But they were TCD Chad deemed sufficiently rigorous to be instructive on UGA Uganda the contours of socioeconomic welfare to provide credible grounds for policy recommendations. The 14 countries emerging from this process The data cover most major displacement events cover nearly all world regions and a large share of from the decade preceding the pandemic. Fragility the global displaced population. Together, these in the Sahel is represented by Burkina Faso and 14 countries hosted more than 25 percent of the Chad. Displacement from conflict in the Levant 79 million people that had been forcibly displaced is represented by Iraq and Jordan. The Rohingya before the start of the pandemic13, and 13 are among crisis is covered by Bangladesh. Displacement the 30 LMICs with the largest counts or shares (or from Venezuela, Nicaragua, and Cuba is shown both) of FDPs in the world14 (see Figures A2.1a and in surveys from Ecuador, Costa Rica, and Mexico. A2.1b in Annex 2).15 These countries also run the Violence in East Africa is reflected in data from IDPs full range of accommodation arrangements, from in Somalia and refugees fleeing to Ethiopia, Djibouti, those that have no or very few UNHCR camps or Kenya, and Uganda. And the simmering conflict settlements, as in Costa Rica, Democratic Republic in Central Africa is evident in data from the DRC of Congo (DRC), Ecuador, and Mexico, to Burkina on two displacement groups: refugees and IDPs. Faso and Djibouti, where nearly half are in camps, Together this collection of surveys covers refugees to Bangladesh, Chad, Ethiopia, and Uganda, in 12 countries and IDPs in four and includes data where nearly all displaced persons are in camps or on camp status in seven countries, as seen in Table settlements (see Table 2.1). 2.2.16 12  “Camps” refers generally to formal or informal camps, settlements, or in the case of Djibouti, “refugee villages.” 13  “Refugee Data Finder,” https://www.unhcr.org/refugee-statistics/ Because some surveys only measure IDPs or refugees even if a country has both, this set of surveys represents about 20% all FDP populations. 14  The exception is Mexico, which hosts the 32nd highest number of FDPs at 443,000 and is the 53rd highest as a share of the national population (0.4 percent). 15  “Refugee Data Finder,” https://www.unhcr.org/refugee-statistics/; and “Macro Poverty Outlook,” https://www.worldbank.org/en/ publication/macro-poverty-outlook. 16  The last column in Table 2.2, labeled “Camp/Non-Camp Data,” indicates only the countries where information on in-camp or out-of-camp status of displaced populations was collected in the HFPS data. A missing mark could indicate that there are no camps in the country (as is the case in Latin America, including Costa Rica, Ecuador, and Mexico, see Table 2.1), information on accommodation type is not being collected, or the sample is being restricted to either only camped or non-camped populations. For example, nearly all refugees live in camps in Chad, whereas no camps exist for refugees in Uganda. In some analyses, the sample size of those in camps may not be large enough to generate reliable statistics (typically with a sample size <30). 7 Table 2.1 Accommodation in Camps Share Living in Camps and Country Total IDPs or Refugees (2019) Settlements (2021) BFA (IDPs) 560,000 52% BGD (refugees) 854,813 100% CRI (refugees) 114,186 0% DJI (refugees) 30,792 50% DRC (IDPs) 5,512,000 5% DRC (refugees) 526,925 25% ECU (refugees) 503,607 0% ETH (refugees) 734,800 91% IRQ (IDPs) 1,555,000 15% IRQ (refugees) 286,924 36% JOR (refugees) 744,951 17% KEN (refugees) 489,728 84% MEX (refugees) 150,950 0% SOM (IDPs) 2,648,000 41% TCD (refugees) 446,426 86% UGA (refugees) 1,381,116 99% Sources: Staff calculation based on CONASUR and UNHCR (2021); OCHA (2021); UNHCR and CNR (2021); CCCM Cluster Somalia (2021); Uganda (2021); UNHCR and Bangladesh (2021); UNHCR (2021a, 2021c, 2021d, 2021e, 2021f); CCCM, REACH, and UNHCR (2021); and “Refugee Data Finder,” https://www.unhcr.org/refugee-statistics/. Table 2.2 Available Samples in the Harmonized Data Country Host Refugees IDPs Camp/Non-Camp Data Bangladesh X X Burkina Faso X X X Chad X X X Congo, Dem. Rep. X X X Costa Rica X X Djibouti X X X Ecuador X X Ethiopia X X X Iraq X X X X Jordan X X X Kenya X X X Mexico X X Somalia X X Uganda X X Source: Staff illustration based on the harmonized HFPS database. 8 2.2  Phone Survey Sampling Strategies are persons of concern (PoCs), including refugees and IDPs in UNHCR field operations. It is now a among Displaced Populations key instrument for the delivery and tracking of protection and assistance services to PoCs around The representativeness of the HFPS data is the world. Coverage of FDP types in ProGres determined by the availability of a comprehensive varies by context. Verification exercises are sampling frame. This list (or other device) can be carried out periodically to update the information linked to contact information for the universe of in each country-specific registration database—a households from which a sample is to be drawn. necessary process for populations as dynamic as FDPs. Importantly, ProGres includes phone Sampling strategies are adapted to the context, numbers and basic demographic characteristics population, and availability of a frame. Three that can be used to sample and contact displaced primary approaches dominate the 22 different populations. The database was used in several sampling strategies of the 30 population groups contexts, either directly by UNHCR or through used in this report. Table 2.3 briefly describes the data-sharing agreements with the World Bank, survey samples for the phone surveys of the 14 depending on which institution led the HFPS data countries harmonized in this endeavor, and more collection exercise on the displaced sample. detailed descriptions are available in the country tables in Annex 1. Sampling based on Random Digit Preexisting samples can be useful in constructing Dialing (RDD) was used in six instances, sampling high-frequency phone panels of displaced and frames based on preexisting surveys were leveraged host populations. Bangladesh, Chad, and Djibouti in nine cases, and registration and population were the only countries where refugees had administrative databases were employed in 13 already been integrated into previous surveys, and others.17 Some surveys employed multiple sampling only in Chad were refugees included in a national frames or approaches, as described in the examples household survey. That lack of inclusion in standard of Kenya and Bangladesh in Box 2.1. 18 national household surveys undermines the principles of statistical inclusion and severely limits UNHCR’s Profile Global Registration System the ability to do rapid, representative survey work (ProGres) database can be a powerful resource in when sudden needs arise or to perform longitudinal sampling displaced populations in contexts where analyses in tracking representative cohorts over it is current and complete. ProGres is UNHCR’s time (World Bank 2023d). main repository for storing data on individuals who Box 2.1 Use of Multiple Sampling Frames in Kenya and Bangladesh The rapid response phone survey (RRPS) sample in Kenya aimed to be representative of refugees and stateless people registered by UNHCR by leveraging the most recent data available for each of five strata—Kakuma refugee camp, Kalobeyei settlement, Dadaab refugee camp, urban refugees, and Shona stateless people. For refugees in Kakuma and Kalobeyei, as well as for stateless people, recently conducted socioeconomic surveys (SES) were used as sampling frames. Because no recent survey existed for urban or Dadaab refugees, those sampling strata were based on ProGres records. The HFPS in Bangladesh used phone numbers for Rohingya refugees and host populations living in the Cox’s Bazar district collected in the 2019 Cox’s Bazar Panel Survey (CBPS) baseline. However, that survey used multiple frames to generate the sample. The CBPS baseline used satellite imagery, combined with the 2011 Bangladesh census, to draw the host sample and used round 12 of the International Organization for Migration’s Needs and Population Monitoring (NPM) site assessment implemented from August to October 2019 to sample Rohingya refugees.19 17  Detailed discussion of these and other approaches can be found in Himelein et al. (2020). 18  “IOM Bangladesh – Needs and Population Monitoring NPM,” https://data.world/iom/1b88bca6-2d7c-423e-97d7-17160d056e9a. 9 Table 2.3 HFPS Design for Displaced and Host Samples Lead Institution Sample Size Geographical Country for Data Population (First Round Sampling Frame Coverage Collection of Analysis) Cox's Bazar district Host & Bandarban district 1,816 (partial) Cox’s Bazar Panel Bangladesh World Bank Survey baseline (2019)b Camped refugees in Refugees 1,358 Cox’s Bazar Burkina NSO, World Host National 1,998 2018/19 EHCVMb Faso Bank IDPs 9 of 12 regions 1,146 CONASUR databasea NSO, World Host National 1,609 2018/19 ECOSIT4b Chad Bank Refugees 10 regions 919 2018/19 RHCHb World Bank Host National 802 phone list / RDDc Costa Rica UNHCR Refugees National 1,163 UNHCR ProGresa Democratic Host Eastern DRC (Beni, 1,252 SPJ-FSRDC registrya Republic of World Bank Refugee Bunia, Goma, Lubero, 126 SPJ-FSRDC registrya Congo IDP and Komanda) 1,087 SPJ-FSRDC registrya 2017 National social Host Urban areas 1,375 NSO, World registrya Djibouti Bank Djibouti-city and 3 Refugees 564 2019 Refugee surveyb refugee villages phone number range / World Bank Nationals National 958 RDDc Ecuador phones with World Bank VDAs National 269 Venezuelan contactd Host National 2,753 2018/19 ESSb NSO, World Addis Ababa, Sub- Ethiopia ARRA/UNHCR Bank Refugees office Jijiga, Sub- 1,676 registration databasea office Shire World Bank Host National 1,623 2018 MICSb Kurdistan and Northern region Phone numbers from Iraq World Bank IDPs 728 (covering approx. MNOsc,d 85% of IDPs in Iraq) UNHCR Refugee National 1,602 UNHCR ProGresa National Unified World Bank Host National 732 registrya Jordan Syrians in Jordan World Bank Refugees registered with 813 UNHCR ProGresa UNHCR World Bank Host National 4,060 2015/16 KIHBSa, RDDc Urban refugees, Kenya Shona stateless and UNHCR Refugees 1,159 SESb, UNHCR ProGresa camps (Kakuma, Kalobeyei, Dadaab) UNHCR Host National; regions 1,142 RDDc Mexico of settlement of 4 UNHCR Refugees primary PoC groups 1,220 UNHCR ProGresa World Bank Host National 2,063 RDDc Somalia IDPs National 718 RDDc World Bank Host National 2,135 2019/20 UNPSb Uganda Kampala, South-West 2018 UBOS surveyb & UNHCR Refugees 2,010 and West-Nile UNHCR databasea Note: For all surveys, the observation unit is the household. “NSO” indicates involvement of the National Statistical Office. Sampling Frame types: a—registry, b—preexisting survey, c—random digit dial, and d—phone list from mobile network operator. ARRA: Ethiopia Agency for Refugee and Returnee Affairs; CONASUR: Conseil National de Secours d’Urgence et de Réhabilitation (database regularly updated); ECOSIT: Enquête sur la Consommation des Ménages et le Secteur Informel au Tchad; EHCVM: Enquete Harmonisée sur les Conditions de Vie des Ménages; ESS: Ethiopia Socioeconomic Survey; IOM NPM12: International Organization for Migration, Needs and Population Monitoring Round 12 data; KIHBS: Kenya Integrated Household Budget Survey; MICS: Multiple Indicator Cluster Surveys; MNOs: Mobile Network 10 Operators; RHCH: Refugees and Host Communities Household Survey in Chad; SES: Socio Economic Survey; UBOS: Uganda Bureau of Statistics; UNFPA PESS: United Nations Population Fund Population Estimation Survey of Somalia; and UNPS: Uganda National Panel Survey. 2.3 Biases purposes of the registration database and the perceived barriers to registration. Despite best efforts, biases may affect any sample, RDD approaches are not immune to bias. Even if as some parts of the true distribution are over or all households do have phones, households with under-represented. As described in Tanner (2021), phone numbers that fall outside of the numerical phone surveys among displaced populations can range or lists used in RDD algorithms do not have a be particularly challenging. However, bias can be chance of being selected, creating a bias that may significantly reduced by being aware of its potential be difficult to sign. This can happen with a foreign sources and acting to mitigate them through ex country code or prefix of a displaced persons ante design and ex post reweighting strategies to phone number, for example. Similarly, in RDD cases bring the sampled data as close to a distribution of where phone lists are used, minority subgroups of the true population distribution as possible. the population may not be on the list. Though not necessarily a bias, it is also worth noting that RDDs Frame bias can be inefficient if the size of a target subpopulation is small relative to the size of the full population in All phone surveys are susceptible to some the frame. For this reason, RDD is often not applied common elements of frame bias. Frames for phone unless the target subgroup is a sizable share of the surveys are only representative insofar as functional overall population; for example, this is why RDD phone access is distributed uniquely and uniformly is feasible in Somalia, where 17.5 percent of the across the population and screening instruments population is estimated to be internally displaced— can accurately identify targeted subpopulations.19 one of the highest IDP rates in the world. Typically, households with no phones are more likely to be poorer and located in remote areas, which could severely affect the representativeness Modality bias of HFPSs.20 Insofar as displaced populations are Conducting interviews through a phone modality more likely to be poorer or have less discretionary offers the enumerator limited control over space in household budgets, they are less likely respondents’ environment and level of focus to have access to a mobile phone. Consequently, during the surveys, which may affect data quality. analysis of phone survey data could overestimate However, studies of nondisplaced populations the welfare of displaced populations relative to have shown that there is little difference between nondisplaced populations such that the actual answers given through phone surveys and in-person disparity between host and displaced populations interviews (Ballivian et al. 2013; Garlick, Orkin, and may be larger than what is reported in these data. Quinn 2015). Emerging HFPS results from the data used in this report on refugees and hosts in Jordan, Registration databases, previous surveys, for example, suggest that phone surveys and censuses, and other list-based frames may be face-to-face surveys generate similar responses outdated or susceptible to self-selection. If they for most topics, but that refugee respondents are have not been recently refreshed or validated, less likely to acknowledge personally sensitive these frames may miss households that have only challenges, such as mental health and depression, recently entered the population, may lack accurate when answering by phone than when answering in contact information, or may list outdated household person (Rodriguez and Smith, forthcoming). characteristics that would be used to formulate sampling clusters or strata (e.g., dwelling location, The volume and complexity of information that household size, education levels, or marital status). can be retrieved over the phone tend to be rather Households may also have differential incentives to limited compared to the quantity and depths of register themselves depending on the perceived 19 Shared phone numbers, households or individuals with multiple phone numbers, uneven phone ownership, unequal network coverage or network, mobile throttling, unreliable electrification, phone time rationing, and so forth, can all contribute to violating the central assumption in the sampling frame that all respondents have a known and nonzero probability of being successfully contacted. The “coverage gap” remains significant in Africa, where, despite a 21 percent increase in 4G coverage since 2020, 18 percent of the population remains without any access to a mobile broadband network (ITU 2021). The continent is also lagging behind in terms of 4G network coverage, and 30 percent of the rural population has no access to the internet. 20  Face-to-Face approaches can help where possible. See Dabalen et al. (2016). 11 topics surveyed face to face because of time Surveys often employed oversampling strategies restrictions. Ten to 15 minutes is usually the upper in the first round to ensure that the desirable size bound of good practice for phone survey length. of the representative sample estimated through Even so, various approaches can be used to retain power calculation was reached. In expectation an appropriate balance between data richness of non-response and attrition, surveys often drew (breadth and depth) and data quality. For example, larger samples than power calculations indicated questionnaires may rotate modules between would be needed, and then put in place recontact waves such that topics that are subject to change protocols to maximize the likelihood of retaining over time (e.g., employment or school attendance) respondents. Expecting challenges when contacting may be kept across waves of phone surveys, but displaced respondents, at least five (Burkina Faso, topics unlikely to see frequent change (such as Chad, Djibouti, Ethiopia, and Uganda) out of the 14 demographic characteristics of the respondent or country surveys oversampled in the initial survey household) may be rotated out (Tanner 2021).21 round. Non-response and attrition rates of HFPS Ex post bias correction As with all data collection approaches, phone After collection, data can be treated to mitigate surveys encounter difficulties with non- some of the bias from incomplete frames, non- contact, non-response, and attrition; these response or attrition to make the distribution of are particularly challenging among displaced observed characteristics as similar as possible in populations. Households may not be successfully the sample and true population. The post-survey contacted because phone numbers may not weighting strategies used in the 14 countries to work or respondents may not be in range or pick correct for bias include propensity score matching up because they do not recognize the caller or or inverse probability weighting and cell weighting. because they are rationing minutes. Similarly, In Himelein’s (2014) ex post weighting approach, a households may not be successfully interviewed propensity score model is estimated using a logit because of refusals, disconnections, or hang- regression of household characteristics on the ups mid-interview. Moreover, there are generally respondent’s likelihood of having completed the higher non-response and attrition rates among interview in the previous wave. Attrition adjustment those experiencing displacement. This observed factors are derived from the propensity scores and phenomenon could be due to several reasons: applied to the initial propensity score matching their frequent need to move, resulting in imperfect weights. A trimming and imputation procedure or outdated sampling frames; their limited access to is then applied to the right tail of the distribution cell phone ownership and rationed use associated before proceeding to a final post-stratification with poverty; and their settlement in remote areas adjustment to get the final balanced panel weights. with limited network coverage (Tiberti et al. 2021; Ambel, McGee, and Tsegay (2021) have shown that Malaeb et al. 2021; World Bank 2020b, 2020d, reweighting techniques can reduce overall sample 2021a, 2021b`). To reduce non-contact and non- bias in HFPS but do not fully eliminate them.22 response, teams in several countries (e.g., Djibouti, Uganda) sent a text message to mobile phones to 2.4  Harmonization of the HFPS Data determine if the number was still functioning and to alert respondents that they would be receiving Survey instruments and questionnaires a call with a survey. In some instances, top-ups are used to offset time used in the survey and provide The first data policy recommendation in the a (small) incentive to participate. World Development Report 2023 for migrants and refugees is to harmonize data and data 21  Though not used in the HFPS here, a random module assignment paired with imputation methods can also be considered. 22  For example, the scope for correction is more limited when RDD or list-based sampling frames are used because there are few (or no) known characteristics about households that were not part of those frames, or about households that were selected but did not respond (Himelein et al. 2020; Brubaker, Kilic, and Wollburg 2021) 12 collection methodologies from across contexts. as well as the duration of and interval between The work in harmonizing data from across these rounds, varied from country to country as shown in 14 countries highlights the value of following Table 2.4 below. Most of the original datasets are that recommendation. The data collected in the available for public use on the microdata libraries HFPSs covered several topics. The core baseline of the World Bank and UNCHR.24 Most results of questionnaire was designed by the World Bank’s this data harmonization exercise from displaced Poverty and Equity Global Practice to facilitate populations can be found on the World Bank’s international comparison (see Table A2.1 in Annex COVID-19 dashboard.25 2 for the list of modules covered in the core questionnaire and brief descriptions). Challenges addressed by harmonization and analysis Country teams often adapted the questionnaire Most of the challenges encountered in harmonizing to accommodate the needs and contextual this large number of sample populations and differences in each country. Such modifications periods are not new. First, not all country surveys were made regarding the inclusion of modules and harmonized in this exercise were explicitly based questions, the wording of questions, recall period, on the World Bank’s core questionnaire or were or changes to response options. The resulting originally designed as part of the Bank’s COVID-19 heterogeneity in survey instruments created monitoring effort. Even so, surveys administered challenges in comparing data from one context by UNHCR to displaced populations in Mexico, to another, requiring that the data be harmonized Costa Rica, Uganda, and Kenya used the Bank’s to facilitate aggregation, where possible. Not COVID-19 core questionnaire and were designed to all countries that included samples of displaced be identical to the questionnaires administered by populations incorporated all modules or items from the Bank to the national samples in these countries. the core questionnaire that allowed harmonization. However, the surveys in Cox’s Bazar Bangladesh, This report presents results from the harmonized in the Eastern DRC, among Iraqi refugees, and in database for select modules and items from these Somalia were developed as part of separate efforts core modules—specifically examining employment, and used distinct survey instruments. Nonetheless, income, assistance and coping mechanisms, food some specific items from those questionnaires were security, and education. sufficiently similar to the Bank’s HFPS instruments that their data could be harmonized and included. Harmonization process Second, customization led to variation in the list The harmonization of the datasets from these 14 of modules and in the specific variables included countries was conducted according to a global in each module across countries and within dictionary developed by the World Bank.23 The countries over time. As a result, although there same harmonization procedure was applied are 14 countries with surveys considered for this to UNHCR surveys. For each topic of interest, report, coverage of modules and outcomes varies relevant questions were screened for consistency substantially. Differences in survey item recall with the Bank questionnaire, and responses were periods can create challenges. For example, in adjusted as needed to facilitate direct comparison. order to better balance coverage and comparability, Some flexibility in the wording of the question and considering the importance of both, a decision was deemed acceptable. The number of rounds, was made to consider comparable food security– 23  An overview of the global initiative, questionnaire template, and other documents is available at World Bank, “Household Monitoring Systems to Track the Impacts of the COVID-19 Pandemic,” https://www.worldbank.org/en/topic/poverty/brief/high- frequency-monitoring-surveys. The data dictionary and other resources for the Harmonized COVID-19 Household Monitoring Surveys can be found at https:// datacatalog.worldbank.org/int/search/dataset/0037769/harmonized-covid-19-household-monitoring-surveys 24 World Bank, “High-Frequency Phone Surveys,” https://microdata.worldbank.org/index.php/catalog/ hfps/?page=1&ps=15&repo=hfps; and UNHCR, https://microdata.unhcr.org/index.php/home. 25  World Bank, “COVID-19 Household Monitoring Dashboard,” https://www.worldbank.org/en/data/interactive/2020/11/11/covid-19- high-frequency-monitoring-dashboard. 13 Table 2.4 Timing of Phone Surveys 2020 2021 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 Bangladesh Burkina Faso Chad Congo, Dem. Rep. Costa Rica Djibouti Ecuador Ethiopia Iraq Jordan Kenya Mexico Somalia Uganda Note: The survey months represent the month(s) during which the majority of the data collection was conducted. related variables, regardless of the reference harmonized database allows for direct and more period.26 Additionally, the different types of accurate comparisons across the populations sampling frames, geographic coverage areas, and included in the sample but can only do so insofar weighting methods used needed to be understood as sufficiently similar data are collected within each in order to make correct interpretations of the data country. Consequently, harmonized databases and the analytical results. can lack the depth of any of the country datasets. Improved ex ante standardization can significantly Third, the timing and frequency of the survey help to increase the richness of harmonized data. rounds varied greatly. The number of rounds, and For example, a module on income loss might ask the duration and interval between rounds, varied additional questions on the source or relative from country to country as shown in Table 2.4 magnitude of different income sources to provide below. In some cases a survey module was only a more detailed picture of income dynamics, done once, preventing analysis of changes over but while those details would have been used in time; in others, the survey was conducted multiple country-level reporting they were not implemented times but in successive months (e.g., Ecuador, across enough countries to include them in the Ethiopia, Iraq, Chad), and in still others, they were harmonized analysis. more spaced apart (e.g., Bangladesh, Costa Rica). Overall, coverage was lower in 2020 compared to With few exceptions, the displaced respondents 2021, when the majority of countries in the sample in these 14 countries were, on average, slightly had at least one survey round. The pattern of younger, less educated, and more likely to be survey coverage over time has implications for the male than their host counterparts. However, there analytical approaches, as discussed in Section 3.1. were no systematic differences in household size across countries. Of course, there are country- Harmonization can limit the richness of the analysis level exceptions to these trends. For example, that is otherwise feasible within countries. The education levels among adults tended to be higher 26  The reference period for food security questions was typically either seven days or 30 days. The original harmonized database considers questions of different reference periods as strictly different variables. Surveys conducted in the same country used the same recall period for all waves and populations. 14 for host populations nearly everywhere except in of observed outcomes to the prepandemic Ecuador, where Venezuelans were on average or pre-displacement/pre-arrival periods limits better educated than their hosts.27 Similarly, socioeconomic interpretation of the results. Similarly, household sizes for IDPs are significantly larger irregular timing of the frequency and number of than nondisplaced households in Burkina Faso, survey rounds during the pandemic limits the use yet internally displaced and host populations in of more sophisticated econometric techniques. And Somalia are virtually identical. third, data on some important aspects of welfare like nominal consumption, mental health, learning, and The study is still limited in several important ways. early childhood development were not collected (or First, the data do not allow for a robust identification were not collected in sufficient countries to allow for strategy to assert the causal link between the more robust interpretation). Even so, the data from interaction of the pandemic with displacement these surveys usefully describe the contours of status on observed outcomes. Second, although the relative welfare of displaced and nondisplaced baseline data is available in some cases, comparison populations during the pandemic. Table 2.5 Descriptive Statistics for Harmonized Survey Data Respondents Average Sub- Age > No Primary Secondary Tertiary Country Household Male Age Population 25 Education Education Education Education Size Refugee 5.42 0.47 37.03 0.81 Bangladesh National 5.27 0.50 35.73 0.80 IDP 13.55 0.65 45.99 0.95 0.88 0.07 0.05 0.00 Burkina Faso National 7.06 0.80 45.95 0.96 0.73 0.12 0.11 0.04 Refugee 6.14 0.48 42.60 0.92 Chad National 6.63 0.82 40.73 0.92 Refugee 7.18 0.43 32.99 0.77 0.12 0.25 0.56 0.07 Congo, Dem. IDP 7.66 0.50 32.74 0.79 0.07 0.25 0.60 0.08 Rep. Returnee 7.57 0.57 32.60 0.76 0.05 0.23 0.63 0.09 National 7.28 0.53 31.91 0.76 0.03 0.13 0.64 0.21 Refugee 5.79 0.49 0.88 0.18 0.28 0.54 0.00 Costa Rica National 3.56 0.43 42.49 0.87 0.02 0.34 0.40 0.24 Refugee 4.32 0.49 37.64 0.90 Djibouti National 6.22 0.48 42.95 0.88 Refugee 4.45 0.60 39.88 0.96 0.00 0.02 0.52 0.46 Ecuador National 4.23 0.44 41.72 0.87 0.00 0.30 0.47 0.23 Refugee 2.66 0.49 28.69 0.57 0.13 0.28 0.53 0.06 Ethiopia National 5.14 0.73 38.77 0.89 0.50 0.19 0.20 0.11 Refugee 3.90 0.35 0.85 IDP 6.50 0.82 36.40 0.84 0.25 0.51 0.11 0.13 Iraq Returnee 6.81 0.88 39.55 0.94 0.21 0.60 0.10 0.09 National 6.40 0.61 37.52 0.90 0.05 0.25 0.22 0.48 27  See, for example, UNHCR Ecuador Monthly Update July—August 2022, which reports that about 23 percent of Venezuelans are estimated to have a higher education degree (UNHCR 2022d). World Bank (2020a) also reports that Venezuelan refugees are typically well educated, with secondary education or above. 15 Table 2.5 Descriptive Statistics for Harmonized Survey Data Respondents (continued) Average Sub- Age > No Primary Secondary Tertiary Country Household Male Age Population 25 Education Education Education Education Size Refugee 5.71 0.43 38.56 0.88 0.08 0.74 0.00 0.17 Jordan National 8.14 0.44 43.60 0.96 Refugee 5.60 0.52 33.63 0.75 0.25 0.34 0.31 0.09 Kenya National 4.13 0.50 35.06 0.80 0.10 0.32 0.38 0.20 Refugee 5.38 0.55 0.78 0.05 0.39 0.46 0.10 Mexico National 6.97 0.44 0.85 0.01 0.11 0.52 0.35 IDP 6.08 0.49 35.20 0.81 0.47 0.29 0.12 0.12 Somalia National 5.46 0.45 35.15 0.80 0.42 0.29 0.14 0.15 Refugee 5.43 0.47 37.53 0.87 0.22 0.35 0.34 0.09 Uganda National 5.05 0.51 42.58 0.93 0.14 0.54 0.28 0.04 Source: Staff calculation using HFPS. 16 The Welfare of Displaced CHAPTER 3 and Host Populations during the Pandemic Displaced women and children walk back to Plain Savo site, Democratic Republic of Congo. © UNHCR/Hélène Caux, March 2022 17 3.1  Pandemic Context of the pandemic. The final data coverage in the T phone surveys for displaced populations was not sufficiently comprehensive to describe how he impacts of COVID-19 were quite changes in welfare outcomes evolved over time widespread and severe, starting in for all relevant population groups. Thus, a more the early months (Egger et al. 2021; careful approach was needed to present a cross- Bundervoet, Davalos, and Garcia 2021). The social country narrative without compromising country- and economic impacts evolved significantly over level heterogeneities. The presentation of results in the duration of the pandemic; in fact, the impacts this section follows a few principles. When there is during the first few months after the onset in 2020 sufficient coverage with data available in about half looked very different from those in, say, mid-2021 or more countries, cross-country aggregated trends (Brunckhorst et al. 2023). are generally presented first. Those are compared with country-level trends to ensure that they The policy response to the pandemic was broadly reflect the trend for the majority of cases unprecedented and included a number of and are not driven by compositional changes in the measures that restricted population mobility over sample over time. Results from individual countries, an extended period of time. These included, in rather than cross-country averages, are presented particular, non-pharmaceutical responses, such when the outcome of interest is available for only as lockdowns and contact tracing. The countries a small number of countries. In every instance, the studied in this report followed patterns similar focus is on understanding how hosts and displaced to many others around the world, as shown in groups or different types of displaced groups an analysis of Google mobility trends and policy (refugees compared to IDPs, in camp compared to stringency indicators28 in 2020 and 2021 (Figure out of camp) fared differently before and during the A2.2 in Annex 2). The trends suggest three different pandemic. stages of the pandemic: the first period from April to June 2020, coinciding with lockdowns in most countries and an initial shock that led to sharp 3.2 Employment increases in policy stringencies and corresponding decreases in mobility; the second period from July As economic activities contracted around to December 2020, which was associated with a the world, the COVID-19 pandemic led to gradual lifting of lockdowns and other containment unprecedented levels of labor market losses. measures, along with a gradual increase in mobility; The harmonized phone surveys made it possible and the third in 2021, which broadly represented to understand some important aspects of the labor a further relaxation of stringency measures and a market impacts during the pandemic, primarily return to prepandemic mobility patterns. those related to job losses that represent changes to employment in the extensive margin. Changes in The analysis in this report is anchored on these the intensive margin, such as changes to hours or three periods, which each define different stages wages, were often not captured and are therefore 28  Policy stringency indicators are estimated using data from the Oxford COVID-19 Government Response Tracker (OxCGRT), https://github.com/OxCGRT/covid-policy-tracker. 18 Figure 3.1 Share of Employed by Host and FDP Type, before and during the Pandemic (%) Figure 3.1 80 Share of respondents (%) 60 Refugee 40 IDP Hosts 20 0 Prepandemic Apr−Jun Jul−Dec 2021 2020 2020 Source: Staff calculation using HFPS. Note: Prepandemic refers to recall questions asking about the period immediately before the pandemic. Household sample weights are used within countries and each country is weighted equally. Confidence intervals, shown as vertical lines, are based on heteroskedasticity robust standard errors. The countries included in the sample for each period are Bangladesh, Ecuador, Kenya, Somalia (Apr-June 2020); Bangladesh, Djibouti, Ecuador, Ethiopia, Iraq, Kenya, Uganda (Jul-Dec 2020); Bangladesh, Burkina Faso, Chad, Costa Rica, Djibouti, Ecuador, Jordan, Kenya, Mexico, Somalia, Uganda (2021). not included in the global harmonized database. employment trends between hosts and FDPs was A large number of workers were likely temporarily relatively minimal. absent, but the surveys typically did not collect sufficiently detailed information to classify them as FDPs experienced disproportionately larger either employed, unemployed, or out of the labor losses in the majority of countries, but this is force. Given that the surveys were administered after not unexpected given their employment profile. the onset of the pandemic, changes in employment Refugees and IDPs are much more likely to be were measured using a retrospective question working in informal, low-skilled jobs, often as day about labor market engagement immediately prior laborers or other types of irregular work. This to the pandemic or the start of lockdowns. means that they frequently lack de jure or de facto protections and were more likely to lose their jobs Aggregate labor market dynamics suggest that the (Dempster et al. 2020; Vishwanath, Alik-Lagrange, initial employment losses were slightly greater and and Aghabarari 2020). Moreover, because of legal the recovery slower among refugees compared restrictions, shorter time horizons, lack of access to hosts. Employment levels fell sharply in the first to financial markets, and socioemotional trauma, three months of the pandemic, after which a gradual general labor market participation (as well as wages recovery began; however, the initial decline was and working conditions) is often lower among steeper among refugees and their recovery was displaced populations (Schuettler and Caron, slower and remained incomplete as of late 2021, 2020). A simple way to examine the differential where data were available. In contrast, employment labor market impact is by looking at work stoppage levels among hosts had recovered significantly by rates among hosts and FDPs, as shown in Figure the second half of 2020. IDPs’ employment patterns A2.4 (Annex 2). Work stoppages are estimated as appeared to follow more closely those of hosts in the share of respondents who were working before the initial period, although the recovery appeared the pandemic but not at the time of the survey. Of to be slower. Figure 3.1 shows aggregate patterns note, both current employment levels and work based on data from 12 countries, while country- stoppages are relevant to consider, because the level figures are presented in Figure A2.3 (Annex 2). latter are only estimated among those who had The cross-country patterns are generally reflective held a job before the pandemic, whereas there of country-level trends, with a few exceptions: in may have been new labor market entrants after the the three Latin American countries (Costa Rica, pandemic started who would be captured in current Ecuador, and Mexico) and Somalia, the difference in employment estimates.29 29  Labor market entry rates (calculated among those who were not working before the pandemic) were estimated to be on the order of 10–15 percent in Burkina Faso, Costa Rica, and Uganda, for example. 19 Multiple structural and pandemic-related factors a work permit; and (5) places additional restrictions may explain the difference in employment levels in terms of work, such as specifying the industries across countries and population groups. The or locations they may work in.31 probability of work is lower among refugees even after controlling for basic respondent characteristics, As expected, the findings suggest that more including gender, age group, and household size. restrictive work rights are associated with Figure 3.2 shows that the marginal effect of FDP lower levels of employment among refugees.32 type on probability of work during the pandemic This result is based on a multivariate ordinary was negative and statistically significant across the least squares (OLS) regression that controls for majority of countries (regression results are in Table basic individual and household characteristics, A3.1a of Annex 3). Where that is not the case, the in addition to economic conditions (proxied using estimated coefficient is small and not statistically GDP per capita and GDP per capita growth) as well different from zero. Differences in structural as policy stringency during the pandemic (Figure economic conditions and the level of policy 3.4 and Table A3.2). Although this may not be a stringencies during the pandemic likely influenced surprising result, the relationship between labor employment outcomes across countries. These market policies for refugees and their employment factors are captured using GDP per capita, annual outcomes across countries has rarely been GDP growth, and the Oxford policy stringency investigated in an empirical manner. This may be index. As expected, more stringent policies and a partly due to the lack of comprehensive information larger negative GDP shock were correlated with a on the laws and legislation that are relevant to FDPs higher probability of having stopped work during across countries, although several recent efforts the pandemic, while the coefficient on the level of have helped to advance the understanding of this GDP is not statistically significant (Figure 3.3 and issue.33 World Bank (2023c) also recently found Table A3.2). that refugees in countries with more liberal refugee policies had higher employment rates. In addition, differences in refugees’ access to labor rights in host countries are expected to Significant job transitioning took place among explain some of the variation in access to job both host populations and FDPs, but there was opportunities. For this analysis, the Developing more churning among the latter group. In a few World Refugee and Asylum Policy (DWRAP) dataset countries, the phone surveys collected additional was used, which provides information on national information on the sector of activity and employment policies toward displaced populations for 92 type for both current and prepandemic jobs, developing countries between 1952 and 2021.30 The which was used to estimate the share of workers database quantifies and codifies the differences in who changed jobs.34 Transition probabilities are de jure policies consistently across countries. The estimated by comparing the sector and employment extent of the restrictions on refugees’ legal work type of the prepandemic job and the latest job on rights is constructed using Principal Component record in 2021. Estimates show that a significant Analysis (PCA) and the subcomponents that number of transitions took place across sectors and constitute access to employment in the DWRAP different employment types. Across employment database. Specifically, they measure whether types, the proportion of wage earners saw a large law or policy (1) guarantees the right to work; (2) relative decline, as only 69 percent of hosts and guarantees the right to self-employment or to start 56 percent of FDPs maintained wage employment a business; (3) guarantees the right to work in during the pandemic, and the rest were pushed into professional fields; (4) obligates individuals to hold self-employment or out of the labor force (Table 3.1). 30  The DRWAP database is particularly notable given its comprehensive geographic and temporal coverage as well as the broad set of policy dimensions that are recorded. See Blair, Grossman, and Weinstein (2021) for details. 31  A sixth subcomponent relates to taxes and quantifies whether the law or policy obligates individuals to pay taxes. Given the relevance for access to work itself, this particular aspect is not considered for the analysis in this report. 32  The DWRAP database measures de jure work rights, though in some countries, workers may have de facto access to the labor market. 33  In addition to the DWRAP database described in Blair, Grossman, and Weinstein (2021), see also Zetter and Ruaudel (2016) and Ginn et al. (2022). 34  Of note, however, is that job changes are likely still underestimated, considering that both sector and employment type are highly aggregated so that, for example, movements of wage workers within the broad services sector are not captured in these calculations. 20 Figure 3.2 Marginal effect of FDP type on probability of work during the pandemic Figure 3.2 BFA BGD CRI DJI ECU Refugee IDP ETH IRQ JOR KEN MEX SOM TCD UGA −.6 −.4 −.2 0 .2 Marginal e ect on probability of working during the pandemic, for working age population (18−64) Source: Staff calculation using HFPS. Note: Figure shows the estimated coefficient on the dummy variable indicating whether the respondent is a refugee or IDP, measuring the difference in the probability of work relative to the national population. Results are based on multivariate OLS regressions, where the dependent variable is a binary indicator for whether the respondent is working or not. Regressions control for household size, gender, and age group (whether respondent is age 25 and above). Individual country regressions include survey month fixed effects. Standard errors are heteroskedasticity robust. Confidence intervals, shown as horizontal lines, are based on heteroskedasticity robust standard errors. Results are robust to alternative estimation techniques, including probit. Figure 3.3 Marginal effect on probability of stopping work Figure 3.3 Refugee IDP HH size With country and Gender male month fixed e ects Without fixed Age above 25 e ects Stringency Log of GDP per capita GDP per capita growth rate −.05 0 .05 .1 .15 Marginal e ect on probability of stopping work, if working prepandemic Source: Staff calculation using HFPS. Note: Figure shows the coefficient on the dummy variable indicating whether the respondent is a refugee or IDP, measuring the difference in the probability of work relative to the national population. Results are based on multivariate OLS regressions where the dependent variable is a binary indicator for whether the respondent has stopped working during the pandemic. GDP data are from 2020. Regressions control for household size, gender, and age group (whether respondent is age 25 and above). Standard errors are heteroskedasticity robust. Confidence intervals, shown as horizontal lines, are based on heteroskedasticity robust standard errors. Results are robust to alternative estimation techniques, including probit. 21 Figure 3.4 Marginal Effect on Probability of Working during the Pandemic for Refugees Figure 3.4 HH size Gender male Age above 25 Stringency Log of GDP per capita GDP per capita growth rate Restrictiveness of work rights for refugees −.2 −.1 0 .1 .2 Marginal e ect on probability of working during the pandemic, for refugees Source: Staff calculation using HFPS. Note: Figure shows the estimated coefficient on the dummy variable indicating whether the respondent is a refugee or IDP, measuring the difference in the probability of work relative to the national population. Results are based on multivariate OLS regressions where the dependent variable is a binary indicator for whether the respondent is working or not. GDP data are from 2020. Regressions control for household size, gender, and age group (whether respondent is age 25 and above). Standard errors are heteroskedasticity robust. Confidence intervals, shown as horizontal lines, are based on heteroskedasticity robust standard errors. Results are robust to alternative estimation techniques, including probit. Table 3.1 Employment Type Transition Matrix, by Hosts and FDPs Share of Probability of Transition into… (%) Share of Prepandemic Working-Age Working-Age Population Employment Population, Self- Wage Not Population, Type Prepandemic employed Earner working 2021 (%) (%) Self-employed 38% 79% 3% 17% 36% National Wage-earner 25% 8% 69% 22% 21% (hosts) Not working 37% 8% 6% 86% 44% Self-employed 24% 73% 7% 20% 26% FDPs Wage-earner 34% 13% 56% 31% 24% Not working 41% 9% 6% 85% 51% Source: Staff calculation using HFPS. Note: Each number represents the probability of a job transition between different employment types, e.g., from self-employment before the pandemic to wage employment in 2021. Diagonal entries represent the share of workers who maintained the same prepandemic employment type in 2021. In countries with multiple surveys in 2021, the latest data point is selected. Sample includes Burkina Faso, Bangladesh, Djibouti, Ecuador, and Jordan. Among hosts, workers in the agriculture sector It should be pointed out that this is against a were much less likely to have moved, whereas context in which employment among FDPs was among FDPs, the same was true for workers in already of low quality. This is not obvious at first the category of “other services.” FDPs were more glance, as the overall distribution of employment likely to have changed jobs compared to hosts types and sectors shows that FDPs were more across all sectors except other services (Table 3.2). likely to be in wage employment and less likely For example, only 23 percent of hosts working in to be engaged in agriculture compared to hosts, agriculture switched jobs during the pandemic both before and during the pandemic. This result compared to 53 percent among FDPs. Figures 3.5 is somewhat driven by the specific countries that and 3.6 present visual illustrations of the relative are included in the sample; in particular, 70 percent size of flows for hosts and FDPs across sectors and of refugees in Ecuador, 30 percent in Jordan, and employment types, respectively. 35 percent in Bangladesh report being engaged in 22 wage-earning work (in Burkina Faso and Djibouti, Stave and Hillesund 2015). In Ecuador, refugees the estimates are 23 and 14 percent, respectively). tend to be engaged in jobs of lower quality—that The reasons for the high share of wage earners are is, jobs that are highly informal and temporary and highly context-dependent but can be attributed to that provide lower returns, which is notable given the availability of cash-for-work programs in camps that Venezuelans Displaced Abroad in Ecuador provided by international or nongovernmental are highly educated (Olivieri et al. 2020). The shift organizations (NGOs) (as in the case of Bangladesh, toward self-employment and agriculture did not see Davis et al. 2022) or informal, temporary wage help in this regard, as both tend to be dominated jobs in sectors such as construction where entry by low-quality jobs associated with low productivity barriers are low (as in the case of Jordan, see and thus less pay and less stability. Table 3.2 Sector Transition Matrix, by Hosts and FDPs Share of Probability of Transition into… (%) Share of Working-Age Prepandemic Working-Age Population Population, Mining / Other Not Sector Population, Prepandemic Agriculture Manufacturing Commerce Services Working 2021 (%) (%) Agriculture 20% 77% 3% 4% 3% 13% 22% Mining/ 7% 7% 61% 4% 7% 21% 8% Manufacturing National Commerce 4% 2% 3% 55% 12% 27% 8% (hosts) Other 19% 9% 3% 13% 46% 28% 15% services Not working 50% 8% 5% 5% 10% 72% 46% Agriculture 8% 47% 1% 3% 5% 43% 8% Mining/ 11% 2% 51% 3% 4% 40% 8% Manufacturing FDPs Commerce 7% 6% 2% 45% 18% 29% 8% Other 18% 1% 3% 3% 67% 26% 20% services Not working 56% 5% 3% 7% 10% 75% 57% Source: Staff calculation using HFPS. Note: Each number represents the probability of a job transition between different sectors of activity, e.g., from commerce before the pandemic to agriculture in 2021. Diagonal entries represent the share of workers who maintained the same prepandemic sector of activity in 2021. Each row sums to 100 percent. In countries with multiple surveys in 2021, the latest data point was selected. Sample includes Burkina Faso, Bangladesh, Ecuador, Somalia, and Uganda. 23 Figure 3.5 Employment Transitions across Employment Types, by Host and FDPs Figure 3.5 Hosts All FDP groups Self−employed Self−employed Wage−earner Wage−earner Not working Not working February 2020 2021 February 2020 2021 Source: Staff calculation using HFPS. Note: Sample includes Bangladesh, Burkina Faso, Djibouti, Ecuador, and Jordan. Figure 3.6 Employment Transitions across Sector of Activity, by Host and FDPs Figure 3.6 Hosts All FDP groups Agriculture Agriculture Mining/Manuf. Mining/Manuf. Commerce Commerce Other services Other services Not working Not working February 2020 2021 February 2020 2021 Source: Staff calculation using HFPS. Note: Sample includes Bangladesh, Burkina Faso, Ecuador, Jordan, Somalia, and Uganda. 24 3.3 Income labor market disruptions were, although this pattern was less clear in Burkina Faso, where IDPs Income losses can be used as a proxy for broader living in camps appeared to have slightly higher welfare losses as they reflect the combined income losses compared to IDPs outside of camps losses from a range of labor and nonlabor (Figure 3.8).35 income sources and have direct implications for household consumption. While the harmonized Between labor and nonlabor income sources, HFPS database mostly captured employment labor income losses were more much more likely changes in the extensive margin, high informality to occur among both FDPs and hosts, though a in these countries meant that many workers were high share of FDPs also reported a decrease in likely forced to cut back on hours or accept lower assistance. For a handful of countries, there is wages (in the case of wage workers) or were additional information on how different components suffering income losses even with the same hours of income were affected, such as wage income, due to decreased demand (in the case of self- farm and nonfarm business incomes, and social employed). In the absence of detailed information assistance. This distinction is useful because on changes in hours or wages, income losses can FDPs rely heavily on assistance from government be considered a next-best proxy for changes that and nongovernment sources (for example, see would be inclusive of those in the intensive margin. World Bank (2021b) for evidence on Chad; World In addition, the loss in overall income could have Bank (2021c) for Uganda; and UNHCR (2022c) also come from losses in nonlabor income, such for Costa Rica), while at the same time, access to as assistance from the government or NGOs. It income-earning opportunities is also typically lower should be noted, however, that the information in for FDPs, as discussed in the previous section. the harmonized dataset indicates only whether Consistent with Figure 3.7, losses from wage and households reported gains or losses in incomes business incomes were widely reported, much more and does not indicate how much incomes so than incomes from private or public transfers changed. This remains an important limitation to (Figure 3.9). the interpretation of the below results. Results from the regression analysis are consistent Many more households lost income than lost with the above outcomes, in that displaced employment during the pandemic. Figure 3.7 populations were more likely to have lost incomes shows that the share of households that lost income even after controlling for demographic and other during the pandemic was multiple times larger than factors such as work stoppages. Figure 3.10 (and the share that reported having stopped work—in the Table A3.1b in Annex 3) shows the results from a case of refugees in Latin America, between 20 and regression of household income losses, conditional 25 percent lost their livelihoods whereas up to 70 on whether the respondent is a refugee as well as percent lost at least part of their income in countries the respondent’s age, gender, and household size, where data was available. Among hosts, the share and experience with work stoppages since the was slightly lower though still staggeringly high, pandemic. The coefficient on the binary refugee with around half of households reporting income variable is positive and statistically significant for losses. These results are reported for refugee and Costa Rica and Mexico. In the case of Ecuador, the host populations in only three countries, all in Latin coefficient on the refugee dummy was still positive American—Costa Rica, Ecuador, and Mexico—so but smaller and (barely) not statistically significant. the results are reported for each country separately. The result that refugees were more likely to have experienced income losses after controlling for Refugees living outside of camps in Chad and work stoppages may indicate that they were more Ethiopia were more likely to experience income likely to have been affected by losses from earned losses compared to camp refugees. This result income (conditional on remaining employed) or may be somewhat expected, given how extensive losses in social assistance. 35  There are no data points for Burkina Faso, Ethiopia, and Chad in Figure 3.7 because they lack information on total income loss for host populations in the harmonized database. Refugees in Costa Rica, Ecuador, and Mexico all reside out of camps as there are no encampment policies in those countries. 25 Figure 3.8 Figure 3.7 Share of Households with Work Stoppages and Income Losses, by Population Group (%) CRI ECU MEX 100 100 100 Share of households (%) 75 75 75 50 50 50 25 25 25 0 0 0 Hosts Refugees Hosts Refugees Hosts Refugees 2021 2021 2021 Stopped work Total income decreased 95% CI Source: Staff calculation using HFPS. Note: Confidence intervals, shown as vertical lines, are based on heteroskedasticity robust standard errors. Figure 3.8 Figure 3.8 Work Stoppages and Income Losses, by Camp Status BFA ETH TCD 100 100 100 75 75 75 Share of households (%) 50 50 50 25 25 25 0 0 0 Camped Out of camp Camped Out of camp Camped Out of camp IDPs IDPs Refugees Refugees Refugees Refugees 2021 Jul−Dec 2020 2021 Stopped work Total income decreased 95% CI Source: Staff calculation using HFPS. Note: Confidence intervals, shown as vertical lines, are based on heteroskedasticity robust standard errors. Figure 3.9 Share of Households Reporting Income Losses, by Source of Income (%) Figure 3.9 100 90 80 Share of households (%) 70 60 50 40 30 20 10 0 Refugee National Refugee National Refugee Refugee National IDP National Refugee Refugee CRI ECU ETH MEX SOM TCD UGA Agricultural income decreased since pandemic Non-farm business income decreased since pandemic Wage income decreased since pandemic Govt. assistance decreased since pandemic NGO assistance decreased since pandemic Remittances decreased since pandemic Source: Staff calculation using HFPS. Note: In Chad, Ethiopia, and Uganda, this information is only available for refugees in the harmonized database. Data on remittances are not available in these two countries. 26 Figure 3.10 Marginal Effect of Being a Refugee on the Probability of Income Loss during the Pandemic Figure 3.10 CRI ECU MEX 0 .05 .1 .15 .2 .25 Marginal e ect on probability of losing income during the pandemic, controlling for work stoppage Source: Staff calculation using HFPS. Note: Figure shows the coefficient on the dummy variable indicating whether the respondent is a refugee, measuring the difference in the probability of income loss relative to the national population. Results are based on multivariate OLS regressions where the dependent variable is a binary indicator for whether the respondent’s household lost income during the pandemic. Regressions control for household size, gender, and age group (whether respondent is age 25 and above). Individual country regressions include survey month fixed effects. Standard errors are heteroskedasticity robust. Confidence intervals, shown as horizontal lines, are based on heteroskedasticity robust standard errors. Results are robust to alternative estimation techniques, including probit. 3.4  Assistance and Coping households that received assistance are reported for Iraq and Somalia, where there is little difference FDPs were more likely to have received assistance between IDPs and hosts (Figure A2.5 in Annex 2). during the pandemic compared to hosts in nearly Because of insufficient coverage, an aggregated all countries. The main variable in the harmonized trend for IDPs is not shown in Figure 3.11. database related to social assistance coverage is defined as whether households “received any social There was significant heterogeneity in the share assistance since the beginning of the pandemic.”36 of FDPs receiving assistance across countries. In The sources of assistance varied by country but Somalia, the share was low at around 18 percent, generally included cash or in-kind assistance from whereas in Bangladesh, assistance reached the government, NGOs, or other nongovernment almost all displaced households—a necessity as sources. FDPs were about 20 percentage points the Rohingya are restricted from living or working more likely to have received assistance than hosts, outside of camps and the incentive pay received among whom the share receiving assistance through camp-based volunteer programs is very gradually increased from around 18 percent in low (Davis et al. 2022). FDPs in camps generally the first few months of the pandemic to about had a similar or higher likelihood of having received 36 percent in 2021 (Figure 3.11). This pattern is assistance than non-camp FDPs, except in Chad replicated in nearly all countries for which data (Figure A2.6 in Annex 2). are available. Where the estimates are similar between FDPs and hosts (such as in Ecuador), the Although assistance coverage may appear high in difference is not statistically significant. Jordan is certain cases, it should not be taken as evidence an exception in that hosts were more likely to have that FDPs were adequately supported during the received assistance—specifically, 43 percent of pandemic. The question in the HFPS often did not refugees and 78 percent of hosts reported having differentiate between regular assistance that was received support by October 2021, with coverage being provided before the pandemic and additional among hosts likely helped by several emergency relief extended during the pandemic. Another cash assistance programs. Data on the share of IDP limitation is that the harmonized database primarily 36 Typically, in the first wave, households were asked whether households had received any support since the beginning of the pandemic. After that, the question was usually rephrased to ask whether households had received any assistance since the last survey. For the purpose of this report, these measures were combined so that a “cumulative” measure could be consistently created across countries. 27 reports whether any support was provided at all population did not fare better—despite its historic without further details on the duration of assistance scale, support to the poor was often delayed and or the benefit amounts. inadequate (World Bank 2022). This is consistent with the findings in Section 4 that show a decline In reality, the level of support likely fell short of in external aid financing for displaced populations, needs. As seen in Figure 3.9, many households combined with a weak government fiscal response reported that their income from assistance had in these countries. declined. For example, about half of refugees in Costa Rica and Chad, a third of IDPs in Somalia, and The reliance on various coping strategies over 80 percent of refugees in Uganda received further suggests that mitigation measures were less assistance during the pandemic than they insufficient in the face of widespread income did before it started. Further, although about half losses during the pandemic. Figure 3.12 shows the of FDPs received some kind of social assistance share of households that reported having reduced over the first two years of the pandemic in the consumption, drawn down their emergency sample countries, this is much less than the share savings, or sold their assets since the beginning of households reporting income losses—a result of the pandemic. The results vary but are overall confirmed in every country for which there are concerning: in the most extreme case, households data. The high share of households that reported in the DRC reported very high probabilities of income losses (as shown in the previous section) reducing consumption and selling off assets, could indicate the extent to which welfare losses around 60 percent and 40 percent, respectively. went unmitigated by any support provided during The estimate was similarly high for Ecuador, while the crisis. in the rest of the countries, up to 20 percent of households were affected—still a high number. Funding shortfalls were widely and frequently Interestingly, compared to the wide variation across reported by agencies such as UNHCR at the countries, the difference between hosts and FDPs frontline of delivering assistance to the displaced. within countries tended to be rather small most of UNHCR (2021g) notes that the largest area of unmet the time. Although these measures are meaningful, needs during the COVID response was the shortfall they may represent lower bounds in terms of the in cash assistance, followed by access to primary need to cope during the crisis, not least because health care and education. Other agencies such as the poorest households likely do not have many World Food Programme (WFP) similarly reported assets or savings to begin with. In other words, funding gaps amid rising food prices (WFP 2021). less engagement in coping strategies does not The poor and vulnerable groups among the host necessarily imply better welfare. Figure 3.11 Share of Households Receiving Assistance during the Pandemic, by Host and FDPs (%) Figure 3.11 80 Share of households (%) 60 40 Refugee Hosts 20 0 Apr−Jun Jul−Dec 2021 2020 2020 Source: Staff calculation using HFPS. Note: Within-country samples are weighted using household sample weights. All countries are weighted equally. Confidence intervals, shown as vertical lines, are based on heteroskedasticity robust standard errors. Sample includes Bangladesh, Ecuador, and Kenya in April–June 2020; Bangladesh, Ecuador, Ethiopia, Kenya, and Uganda in July–December 2020; and Bangladesh, Chad, DRC, Jordan, Kenya, Mexico, and Uganda in 2021. 28 Figure 3.12 Share of Households that Relied on Various Coping Measures, by Host and FDPs (%) Figure 3.11 80 70 Share of households (%) 60 50 FDPs 40 Hosts 30 20 10 0 Congo, Dem. Rep. Ecuador Ethiopia Mexico Somalia Uganda Ethiopia Somalia Congo, Dem. Rep. Ecuador Mexico Uganda Congo, Dem. Rep. Ethiopia Somalia Reduced consumption (%) Drew on emergency savings (%) Sold assets (%) Source: Staff calculation using HFPS. 3.5  Food Security High prices of food and other commodities contributed to levels of food insecurity not seen in a decade. Indeed, price indices for basic agricultural Going without food is a coping mechanism of products of oils, meals, and grains rose from just last resort. Food security could be considered a under 2010 levels right before the pandemic to reduced form metric of whether households can more than 125 percent higher a year later, and meet minimal welfare needs. Food security during they have stayed high and even increased since the pandemic was not a new challenge for countries that time (Figure 3.13). Following years of declining that host displaced populations. In 2019, just before levels of food insecurity, the risk of hunger and the pandemic shocks hit world food markets, malnutrition shot up during the pandemic and in the more than 80 percent of displaced people were years since due to compounding crises, including sheltering in countries with high levels of acute the recent food shortages caused by the Ukraine food insecurity and malnutrition. Indeed, of the 10 war (see Figure 3.14). For example, areas in Burkina countries hosting the most displaced people, nine Faso faced catastrophic levels of food insecurity in had suffered through a recent major food crisis (WFP 2020 and 2021 (FSIN, 2022). and IOM 2020). During the COVID-19 recession, the situation deteriorated; a joint report by the WFP Food security was an even larger threat for and International Organization for Migration (IOM) those who had been forcibly displaced. Food noted that “no country had been spared” increased price inflation is likely to be particularly regressive; hunger during the pandemic (WFP and IOM 2020). that is, because food forms a larger share of their expenses, higher food prices means that poor In many countries the pandemic disruptions households’ food budgets shrink in real terms. Even were in addition to other, perhaps even more those families that receive some portion of their salient shocks affecting food security. Pandemic- food through in-kind transfers from government or period complications bled into the food price crisis humanitarian organizations are affected by food resulting from Russia’s aggression in Ukraine. Some price inflation. Early in the pandemic, the WFP and nations faced a host of local preexisting, cyclical, or UNHCR reported significant increases in food prices contemporaneously erupting challenges. Yemen, and supply chain challenges (UNHCR and WFP for example, simultaneously experienced currency 2020), making it difficult to bring humanitarian aid to crises and food and fuel price shocks (see Favari the displaced. As illustrated in Section 4, foreign aid et al. 2021; D’Souza et al. 2022). Other countries, was often cut precisely when it was most needed; for including Burkina Faso and Ethiopia, faced lean example, amidst pandemic-induced spikes in food periods depending on agricultural workers’ crops, prices, the WFP was forced to cut food assistance political upheaval, or conflict—all affecting food by 40 percent to the largest refugee population in security during the pandemic (see Rudin-Rush et al. Africa, the 1.3 million refugees in Uganda. 2022, for example). 29 Figure 3.13 Agricultural Price Indices Figure 3.13 180 160 Share of households (%) 140 Oils & Meals 120 Grains Other Food 100 80 60 Jan-20 Mar-20 May-20 Jul-20 Sep-20 Nov-20 Jan-21 Mar-21 May-21 Jul-21 Sep-21 Nov-21 Jan-22 Mar-22 May-22 Jul-22 Sep-22 Nov-22 Jan-23 Mar-23 May-23 Source: World Bank, “Commodity Markets,” https://www.worldbank.org/en/research/commodity-markets#1. Displaced households were more likely to run and country fixed effects. Although there is low on food. In each period of analysis, displaced considerable variation in regressions pooling households were more likely than hosts to run out countries with the necessary data, point estimates of food. This validates concerns voiced by WFP in suggest that refugee and IDP populations are November 2020 that dwindling resources could much more likely to run out of food relative to result in household-level food shortages (UN host populations. This analysis also indicates that 2020). Early in the pandemic, host, refugee, and although household size and age of the respondent IDP populations in the sample were similarly likely (usually the household head) do not seem to be to run out of food, but perhaps because of the correlated with the likelihood of running out of successive shocks to food markets, the share of food, the finance-related covariates—current work displaced households that ran out of food appears status, national GDP per capita, and national GDP to have increased as the pandemic crisis bled growth—are all negatively correlated with running into the food price crisis induced by the Russian out of food due to finances, as expected (Figure aggression in Ukraine. Toward the end of 2021, 3.15 and Table A3.2). nearly 70 percent of forcibly displaced households in the sample reported running out of food because they lacked resources. In fact, in nearly Figure 3.14 Number of People Facing Crisis- every country (except for Ecuador) and in nearly 3.14 Figure of Levels Acute Food Insecurity every period in every country measured (except People facing crisis-levels of acute food insecurity for June 2020 in Somalia), displaced populations 300 were statistically significantly more likely to run out of food than hosts (Figure A2.7 in Annex 2). 250 Running out of food because the household lacked 200 resources was particularly common for displaced Millions 150 populations and hosts in DRC and Chad, but the gap between displaced and host was largest in 100 Uganda. 50 The correlation between being displaced 0 2016 2017 2018 2019 2020 2021 2022 and running out of food is robust, controlling for socioeconomic characteristics, including Source: FSIN, “2023 Global Report on Food Crises” (Rome: Food previous and current employment, food prices, Security Information Network, 2023), https://www.fsinplatform. engagement in the agriculture sector, and time org/sites/default/files/resources/files/GRFC2023-brief-EN.pdf. 30 Figure 3.15 Estimates from the Linear Probability Model on the Likelihood of Running out of Food Figure 3.15 Refugee IDP Working pre−pandemic Currently working HH size With country and month fixed e ects Gender male Age above 25 Without fixed e ects Agriculture sector Stringency Food price inflation (%) Log of GDP per capita GDP per capita growth rate −1 −.5 0 .5 1 Marginal e ect on probability of household running out of food because of lack of resources Source: Staff estimates using HFPS Note: Within-country samples are weighted using household sample weights. Countries are weighted equally. Results from multivariate OLS (linear probability model) regression, with confidence intervals based on standard errors clustered by country. Sample includes Chad, Costa Rica, Ecuador, Mexico, and Uganda. Confidence intervals are based on heteroskedasticity robust standard errors. Results are robust to alternative estimation techniques, including probit. As households run low on food, ration reductions Most concerning, households facing more acute result in displaced household members— food insecurity may have to go an entire day (or sometimes even children—being more likely to longer) without eating. Country-level dynamics skip a meal entirely. Indeed, in each pandemic show that displaced groups were more likely to have period, displaced households were 50–100 gone a day without food in at least one period for all percent more likely to have had an adult skip a 10 countries for which there are data. Moreover, in meal because of a lack of resources (see Figure only three—Djibouti, Ecuador and Somalia—did host 3.16). Childhood malnutrition has been linked to and displaced population levels converge in the last compromised cognition, impaired behaviors, and survey period (see Figure A2.8 in Annexes 2). underperformance at school (see, for example, Martins et al. 2011; Kirolos et al. 2022). The set of As shown in Figure 3.17, on average adults in surveys in this report typically did not collect this displaced households were far more likely in data, but it was collected in Djibouti, Jordan, and every pandemic period to go hungry.37 Even after Kenya. In Djibouti at the end of 2020, children from controlling for household-level socioeconomic urban refugee families were 18 percentage points status, all countries for which data exist, except less likely to have had three meals per day, and Djibouti and Ecuador, showed a statistically significant they were more likely to have gone to bed hungry. higher likelihood of displaced households going a Similarly, surveys in Jordan that generated Food day without food relative to host households. This Consumption Scores indicated that refugee children has near and longer-term effects on labor market were nearly 10 percentage points more likely to income by reducing productivity and may also have have been food insecure during the pandemic than even longer-term effects on household consumption Jordanian children—though the latter were more by shortening working life and potentially increasing likely to have gone to bed hungry or skipped a health care costs. Again, in pooled regression the meal (Rodriguez and Smith, forthcoming). Although finance-related covariates—current work, national this unfortunate gap decreased in Kenya over the GDP, and national economic growth— are seen to course of the pandemic, it was still more common be negatively correlated with going a day or more among refugees than hosts. without food, one of the most severe forms of coping (see Figure 3.18 and Table A3.1c). 37  The dip in the share of host populations not having eaten for a day during the middle of the pandemic is likely because Somalia, which has a relatively high share of all populations going full days without a meal, was absent from that period because data were only collected right at the shoulder of the other two periods in June 2020 and January 2021. 31 Figure 3.16 Share 3.16 of Households with Adults Having Skipped a Meal Because of Lack of Resources (%) 80 Share of households (%) 60 All FDP groups 40 Hosts 20 0 Apr−Jun Jul−Dec 2021 2020 2020 Source: Staff estimates using HFPS. Note: Within-country samples are weighted using household sample weights. Countries are weighted equally. Sample includes Ecuador and Kenya (April–June 2020); Ecuador, Kenya, and Uganda (July–December 2020), and Bangladesh, Burkina Faso, Chad, Djibouti, Kenya, Mexico, Somalia, and Uganda (2021). Household sample weights are used within countries, and each country is weighted equally. Confidence intervals, shown as vertical lines, are based on heteroskedasticity robust standard errors. Results are robust to alternative estimation techniques, including probit. 3.17 Share of Households with Members Not Having Eaten for a Day Because of Lack of Figure 3.17 Figure Resources (%) 80 Share of households (%) 60 40 Refugee IDP Hosts 20 0 Apr−Jun Jul−Dec 2021 2020 2020 Source: Staff estimates using HFPS. Note: Within-country samples are weighted using household sample weights. Countries are weighted equally. Sample includes Ecuador, Kenya, and Somalia (April–June 2020); Ecuador, Kenya, and Uganda (July–December 2020); and Burkina Faso, Chad, DRC, Costa Rica, Djibouti, Ecuador, Kenya, Mexico, Somalia, and Uganda (2021). Household sample weights are used within countries and each country is weighted equally. Confidence intervals, shown as vertical lines, are based on heteroskedasticity robust standard errors. Results are robust to alternative estimation techniques, including probit. Aggregate measures, such as the Food Insecurity severely food insecure. Similarly, in Burkina Faso Experience Score (FIES), accentuate the severity between May and July of 2021, internally displaced of hunger challenges among the displaced Burkinabe were nearly 2.5 times more likely to have during the pandemic. Between January and experienced moderate or severe food insecurity: 74 April 2021, 77 percent of Chadians experienced percent of IDPs experienced moderate or severe moderate or severe food insecurity. As alarming food insecurity compared to 33 percent of their as that number is, over the same period, fully 96 nondisplaced compatriots.38 Aggregating across all percent of refugees in Chad were moderately or pandemic waves, these countries, with the other 38  For further analysis on this data in Burkina Faso and Chad, see Joint Data Center, “JDC Support to Integrating Forcibly Displaced Populations into COVID-19 High Frequency Phone Surveys,” https://www.jointdatacenter.org/jdc-covid-19-hfps/; and Baradine et al. (2021). 32 Figure 3.18 Marginal Probability of Displaced Households with Members Not Having Eaten for a Day Because of Lack of Resources, Relative to Host Households Figure 3.18 BFA CRI DJI ECU KEN Refugee IDP MEX SOM TCD UGA −.2 0 .2 .4 .6 Marginal e ect on probability of household members not eating for a day because of lack of resources Source: Staff estimates using HFPS. Note: Within-country samples are weighted using household sample weights. Results from multivariate OLS regression, with confidence intervals based on standard errors clustered by country. Confidence intervals, shown as horizontal lines, are based on heteroskedasticity robust standard errors. Results are robust to alternative estimation techniques, including probit. two nations in the dataset for which FIES scores can The higher levels of food insecurity experienced by be calculated,39 reveal that in the Sahel, refugees in displaced populations is remarkably consistent, Chad and Uganda appear far more likely than IDPs in though there are some important exceptions. Burkina Faso and Somalia to have been moderately Although only four countries’ surveys included all and severely food insecure, and both FDP groups eight questions that allow for calculation of the were much more likely than host populations to FIES, another eight countries asked questions on have experienced food insecurity (see Figure 3.19). at least two of the eight FIES subcomponents. Averaging all waves within a country, it can be Figure 3.19 Food Insecurity Experience Scores seen that in almost every instance, displaced Figure 3.19 populations fared worse than hosts (Table 3.3). The data that are available also reflect the severe 94.5% 100 vulnerability of some populations, like refugees in 75.1% Uganda, or some regions for both displaced and 80 host populations, like Chad and Eastern DRC. It Share of households (%) is also apparent that even though households 60 52.8% 88.4% 47.3% in Ecuador were relatively more food secure, the higher education level of Venezuelans was 40 37.9% insufficient to insulate them from dimensions of food insecurity relative to Ecuadorian hosts. 20 27.8% 14.9% 6.1% Somalia is the single exception to the trend that the 0 Refugee IDP Host displaced were at least as likely as nondisplaced Moderate food insecurity hosts to encounter every dimension of food Severe food insecurity insecurity (Table 3.3). In Somalia, the internally Source: Staff estimates using HFPS. displaced and nondisplaced populations were Note: In this figure moderate food insecurity refers to households virtually identical in the average number of FIES that experience three or four of the scenarios described in the eight FIES questions, and severe food insecurity refers to subcomponents that respondents affirmed they had households experiencing five or more. Vertical brackets are 95 percent confidence intervals. Sample includes Burkina Faso, Chad, Somalia, and Uganda. 39  FIES scores can be calculated only when a survey includes all eight FIES component questions. 33 34 Table 3.3 Food Insecurity Components Hungry Worried Unable to Average number Hungry but HH ran Ate only a Ate less than adults went about not eat healthy/ Skipped a of affirmative Country Population could not out of few kinds you thought w/o eating having preferred meal subcomponents eat food of foods you should for a day enough food food (1 to 8) Refugee 75% 45% 31% Bangladesh Hosts 73% 44% 32% IDP 25% 15% 27% 86% 74% 78% 38% 58% 4,00 Burkina Faso Hosts 6% 2% 4% 51% 35% 49% 12% 23% 1,83 Refugee 80% 52% 81% 94% 94% 93% 91% 92% 6,77 Chad Hosts 57% 35% 60% 90% 87% 72% 69% 77% 5,47 Refugee 88% 84% 92% 90% Congo, Dem. IDP 85% 79% 91% 91% Rep. Hosts 75% 69% 83% 82% Refugee 33% 68% Costa Rica Hosts 6% 23% Refugee 8% 19% Djibouti Hosts 6% 8% Refugee 48% 15% 47% 42% 56% Ecuador Hosts 27% 11% 39% 42% 33% IDP 64% 33% Iraq Hosts 28% 18% Refugee 32% 15% 55% 33% Kenya Hosts 40% 5% 55% 31% Refugee 55% 36% 60% 58% Mexico Hosts 17% 7% 22% 20% IDP 45% 32% 79% 60% 63% 65% 54% 63% 4,61 Somalia Hosts 48% 36% 66% 59% 64% 65% 61% 64% 4,62 Refugee 84% 43% 72% 92% 93% 93% 87% 84% 6,50 Uganda Hosts 10% 3% 9% 28% 37% 38% 14% 21% 1,59 Source: Staff estimates using HFPS. experienced.40 Of the eight countries globally with pandemic mostly reflected regional trends but more than 10 percent of the population internally was overall characterized by widespread school displaced, only two collected phone survey data closures (Figure 3.21). Schools in Ecuador were during the pandemic: Somalia, in which 17.5 percent closed, partially or fully, for the longest period of is internally displaced, and Yemen which suffers an time, reaching 91 weeks over nearly two years; internal displacement rate of 13 percent. Reports Uganda, Bangladesh, and the other two Latin using the Yemen phone survey data indicate that, as American countries in the sample were not far with Somalia, declines in food security were nearly behind. identical between displaced and nondisplaced Yemenis (Favari et al. 2020). Following widespread school closures, many governments responded by shifting education Households living in camps were not necessarily online, but the modality and implementation less food insecure than those living out of camps. details varied widely. LICs were less likely to Over three rounds of data in Burkina Faso, IDPs offer any distance learning than MICs, and when it in camps consistently had a slightly higher FIES was offered, online learning was available in less than those that are not in camps (though both than half of LICs. Instead, education was typically were far higher than the food insecurity scores delivered using TV and radio in those countries of nondisplaced households) (Tiberti et al. 2021). (Figure 3.22). This is in line with the observation noted earlier for Kenya, wherein nearly every survey period showed Differences in the provision and uptake of learning that children in camped households were more modalities likely reflect the reality in those likely to have skipped a meal than refugee children countries where limited access to the internet, in households living outside camps. electricity, and digital devices made it particularly challenging to participate in distance learning. 3.6  Education and Learning For example, 89 percent of children did not have a computer and 82 percent did not have internet The disruption to education during the pandemic access in 2020 in sub-Saharan Africa (World Bank was historic. By some estimates, over 1.5 billion and UNHCR 2021). In Jordan, only about 2 percent students around the world were affected by of refugee households owned a computer before school closures that were still in place in many the pandemic (Wagner and Hine 2021), and even countries even in late 2021, nearly two years into the simplest technologies such as radios are often the pandemic.41 However, the education response not available to displaced populations (UNHCR differed widely across regions, as seen in the 2020). The displaced are often not connected to UNESCO database on COVID school closures, power or it may not be affordable; over 80 percent which tracked partial or full school shutdowns of refugee camps are estimated to have minimal between February 2020 and March 2022. South access to energy, and where it is available, it Asian countries had the longest closures on can be very expensive.42 These barriers further average, with schools being closed for nearly 70 exacerbated the many challenges that refugee weeks over the 95 weeks of this report period. children faced with regard to accessing education Europe and Central Asia had the shortest closures, even before the pandemic, including documentation but still averaged nearly 30 weeks (Figure 3.20). requirements to enter the host country’s education For children residing in the 14 countries examined system and the lack of access to properly trained in this report, the learning experience during the teachers. 40  Of the eight FIES subcomponents, internally displaced and nondisplaced populations were statistically significantly different for only two: IDP households were more likely to have run out of food, and host households were more likely to have skipped a meal. 41  UNESCO, ”Education: From School Closure to Recovery,” https://www.unesco.org/en/covid-19/education-response. 42  In Dadaab camp in Kenya, refugees spent nearly a quarter of their income on energy before the pandemic (Lahn and Grafham 2015). Still, there are also exceptions: in Jordan’s Azraq camp, electricity generated from a solar plant is being provided free of charge. See C. Dunmore, “Jordan’s Azraq Becomes World’s First Clean Energy Refugee Camp,” UNHCR Stories, May 17, 2017, https://www. unhcr.org/news/stories/jordans-azraq-becomes-worlds-first-clean-energy-refugee-camp. 35 Figure 3.20 Length of Full or Partial School Closure, Average Number of Weeks by Region Figure 3.21 100 No. of weeks 34.1 32.3 24.1 42.2 13.4 18.1 35.4 15.7 29.6 24.8 15.5 18.4 12.8 7.0 0 South Asia Latin America North America Middle East East Asia Sub−Saharan Europe & & Caribbean & North Africa & Pacific Africa Central Asia Fully closed Partially closed Source: Staff calculation using UNESCO, “Dashboards on the Global Monitoring of School Closures Caused by the COVID-19 Pandemic, https://covid19.uis.unesco.org/global-monitoring-school-closures-covid19/. Figure 3.21 Length of Full or Partial School Closure between February 2020 and March 2022 by Country Figure 3.22 100 23 23 28 No. of weeks 51 27 39 10 41 19 66 63 9 53 9 51 5 40 43 44 6 32 28 21 24 23 7 18 19 4 9 7 0 ECU UGA BGD CRI MEX IRQ ETH JOR RWA KEN COD TCD NGA SOM BFA DJI Fully closed Partially closed Source: Staff calculation using “Dashboards on the Global Monitoring of School Closures,” https://covid19.uis.unesco.org/global- monitoring-school-closures-covid19/. Figure Figure 3.22 Distance Learning Modalities Adopted, by Country Income Group 3.23 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Any distance Online TV Radio learning Modality LICs LMICs UMICs Source: “Dashboards on the Global Monitoring of School Closures,” https://covid19.uis.unesco.org/global-monitoring-school-closures- covid19/. Note: UMIC = upper-middle-income countries. 36 These challenges are reflected in HFPS data latest, most comprehensive enrollment data come that reveal large differences in the extent to from UNHCR (2022a), which compiled education which children’s learning was interrupted statistics for more than 40 countries. Average gross during the pandemic. Figure 3.23 presents the enrollment rates at the primary and secondary share of households with children who stopped levels were estimated at 68 percent and 37 learning during the pandemic by country. The percent, respectively, while tertiary enrollment was question is directed to respondents with school- only 6 percent. These estimates are all well below age children, asking whether they were in school the enrollment rates for nondisplaced populations, before the pandemic and whether they had estimated at 91, 84, and 37 percent at the primary, stopped learning once the pandemic began.43 secondary, and tertiary levels, (World Bank and Households in Ethiopia, Iraq, Kenya, and Chad UNHCR 2021). (especially when asked in the first round) reported very high levels of learning interruptions, at Lower prepandemic enrollment rates among around 80 percent or higher. Hosts and displaced displaced children compared to host children populations were affected in similar ways, which were also observed in the HFPS in most countries. is perhaps unsurprising, given that school closure This can be seen in Figure 3.24 which is useful policies in camps reportedly followed national to understanding the average enrollment gap policies during the pandemic (UNHCR 2022e).44 between children from displaced and nondisplaced The impact was much lower in other countries, households. The aggregate figure, however, such as Ecuador, Costa Rica, and Mexico, where conceals wide variation across countries in displaced less than 10 percent of households reported children’s access to schooling (presented in Figure that their children had stopped learning entirely A2.9, Annex 2).46 There were significant differences (Figure 3.23), which may be due to the fact that in displaced children’s prepandemic enrollment despite long school closures overall, schools rates across the countries in the HFPS sample, were partially open for a good number of weeks, ranging from less than 10 percent in Ethiopia to 93 allowing for relatively fewer interruptions to percent in Ecuador (closely followed by Jordan, learning. Learning disruptions were also relatively where enrollment was 86 percent). Prepandemic moderate in Burkina Faso and Jordan, although enrollment rates in the remaining countries were this may have as much to do with the later timing between 60 and 80 percent among children from of the survey—mid to late 2021 in both countries, refugee households in Costa Rica, Kenya, Mexico, when policy stringency was trending toward much Chad, and Uganda. Ecuador is a notable exception lower levels (Figure A2.2, Annex 2). in that refugee children were reported to have been in school before the pandemic or to have continued Refugee children’s access to education was a learning during the crisis at rates comparable to large challenge even before the pandemic.45 The those of host children. 43  Since phone surveys are typically designed to be short, in most cases they did not collect schooling information on each individual child separately, which may contribute to some differences in official enrollment statistics. Household-level results are likely to miss some variations by schooling level (primary, secondary, etc.) and the gender of the child, which limits the understanding of whether boys or girls were more likely to have lost out on learning, for example. In fact, multiple sources raised concerns over girls dropping out of school in higher proportions compared to boys to take on care responsibilities and support income generation, in some cases leading to early marriage (Wagner and Hine 2021), though this appears to also have been true for adolescent boys from disadvantaged families (UNHCR 2022a). 44  This was also verified independently with UNHCR operations for the countries in the sample. 45  Accurate and comprehensive enrollment statistics are difficult to collect. Age-specific enrollment data by international protection status are particularly difficult to collect in many countries (UNHCR 2020). Estimates tend to be more reliable for refugee children in camps where it is easier to collect data, though missing information on their protection status makes it challenging to determine the education status of refugee children who are integrated into national systems and attending public schools (UNHCR 2016, 2022a). 46  The question on access to education before the pandemic was measured quite consistently across countries, asking mainly whether children in the household were enrolled in school or attending school. The question on learning after the pandemic was broader and varied somewhat across countries, as the surveys attempted to capture the new modalities and forms of learning initiated during the pandemic. For example, some surveys asked about continued learning activities, whereas others asked about school attendance (the differences may be related to the presence of lockdowns at the time of the survey). In some countries, this question was asked only when children were in school before the pandemic. In such cases, it is assumed that those who were not in school before the pandemic were not participating in pandemic learning after it got underway. 37 Figure 3.23 Figure 3.23 Share of Households with Children Who Stopped Learning during the Pandemic, by Country (%) Burkina Faso Chad Costa Rica 100 100 100 50 50 50 0 0 0 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Ecuador Ethiopia Iraq 100 100 100 50 50 50 0 0 0 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Jordan Kenya Mexico 100 100 100 50 50 50 0 0 0 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Somalia Uganda 100 100 Refugee 50 50 IDP Hosts 0 0 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Source: Staff calculation using HFPS. Note: Vertical lines represent confidence intervals based on heteroskedasticity robust standard errors. 38 Figure 3.24 Share of Households with Children Accessing Education before and during the Pandemic, by Type (%) FDP3.25 Figure 80 Share of households (%) 60 Refugee IDP Hosts 40 20 0 Prepandemic Apr−Jun Jul−Dec 2021 2020 2020 Source: Staff calculation using HFPS. Note: Prepandemic refers to recall questions asking about the period immediately before the pandemic. Household sample weights are used within countries, and each country is weighted equally. Confidence intervals, shown as vertical lines, are based on heteroskedasticity robust standard errors. The countries included in the sample for each period are Ecuador and Somalia (April–June 2020); Ecuador, Ethiopia, Iraq, Kenya, and Uganda (July–December 2020); and Burkina Faso, Chad, DRC, Costa Rica, Ecuador, Jordan, Kenya, Mexico, Somalia, and Uganda (2021). The data point for refugees in April–June 2020 is omitted due to low coverage. IDPs have their own challenges, even though they The impact of school closures on learning access technically reside in their own country. Children during the pandemic was very heterogeneous. from IDP households displayed enrollment rates This was also true when prepandemic enrollment of around 65 percent in Burkina Faso and Somalia rates were relatively high. Some countries, such as and 77 percent in Iraq, which was lower than that Jordan, Ecuador, and Burkina Faso, experienced of hosts. Among the challenges for IDPs are the relatively few disruptions, whereas in Iraq, Chad, formal registration requirements: for example, and Kenya, learning was interrupted for the vast in Iraq, children can only register for schooling majority of children (Figure 3.25). In Chad and Iraq, during the first 50 days of the school year and can the return to school appears to have been swift be refused registration if they are without identity following the reopening of schools (Figure A2.9, and schooling documentation (IOM 2020), which Annex 2). In Ethiopia, where the baseline enrollment families may inadvertently leave behind in the rate was low, nearly 80 percent of host and chaos of being displaced. displaced children stopped learning once schools closed. For refugee children, access to learning fell Children in households that live in camps tend further from an already low schooling rate before to have better education access than those who the pandemic. Schools reopened in October 2020 live out of camps. In a handful of countries, the but the return to school was slow, possibly because information on access to schooling in the HFPS can many schools did not have the resources to prevent be disaggregated for displaced households living in the spread of COVID-19 (Wagner and Hine 2021). and out of camps. Children living in camps had higher enrollment rates compared to children living out of The heterogeneity in learning is a function of camps in five out of six countries where prepandemic a variety of factors. Policy support for refugee school enrollment rates could be disaggregated education, the effectiveness of educational by camp status (Figure A2.10 in Annex 2). These responses to COVID, and households’ ability to trends may reflect generally higher levels of service access alternative forms of learning offered during provision for displaced populations living in camps. the pandemic likely all contribute to a child’s 39 Figure 3.25 Prepandemic Schooling vs Stopped The inclusiveness of formal education policies Learning during the Pandemic, by Country (%) does not correlate strongly with access to Figure 3.25 schooling before and during the pandemic. For example, Jordan and Ecuador have a policy score 100 IRQ of zero across all policy areas, but both provided high levels of access to schooling in practice. There 80 ETH Stopped learning during pandemic (%) TCD have also been recent efforts by the Jordanian government to include refugees in national KEN education systems (UNHCR 2022b). Chad which 60 recorded a de jure education policy score of four and relatively high levels of education access has UGA been in the process of integrating refugees into 40 SOM their national education system by implementing the 2030 Strategy for Refugee Education that the 20 BFA government signed in 2020. Meanwhile, Ethiopia JOR MEX CRI scores well on the formulation of formal education ECU policies, but de facto access to schooling remained 0 0 20 40 60 80 100 quite low among refugees (Figure 3.26). Schooling before pandemic (%) Access to quality education for refugees is Source: Staff calculation using HFPS. determined by not only education policies but Note: Share of children who stopped learning is based on the earliest available estimate for each country. Countries below the a number of other policies as well, making it a 45-degree line had lower school enrollment after the pandemic started, whereas the opposite is true in countries above the complex challenge to tackle. In many hosting same line. countries where there is ongoing conflict and violence, damage to school infrastructure, a chronic ability to enroll before the pandemic and learn shortage of teachers, and continued security issues during it.47 The inclusiveness of education policies make accessing schooling even more challenging is measured by ranking the formal policy stance (UNESCO 2019; IOM 2020). Refugees also tend to toward refugee children’s education as written settle in poorer areas of the country where education in national laws or legislation. Drawing on the services are of even lower quality (World Bank and DWRAP database, the outcomes in the HFPS data UNHCR 2021) and financial constraints often make are compared against an aggregate sum of two it challenging to keep children in school. Improving components of education policies, namely, whether education outcomes requires investments in the law or policy guarantee access to primary and infrastructure, teachers, outreach efforts, and more, secondary education. Each component takes on a which can be difficult in capacity- and resource- score of zero, one, or two. For simplicity, the scores constrained countries that are already struggling to from the two components are added up to vary improve enrollment among host children. between zero and four.48 Higher scores represent more open, inclusive policies toward displaced populations. The resulting education policy scores range from zero (Ecuador, Jordan) to one (Mexico), and four (Chad, Costa Rica, Ethiopia, Kenya, and Uganda); none of the countries in this report had a score of two or three. 47  In some countries, the question on learning after the pandemic is asked to all households with school-age children, and in other countries it is asked only to households that were sending their children to school before it began. In order to generate comparable estimates across countries, it is assumed in the latter set of countries that if the household was not sending their children to school before the pandemic, they did not do so during it either. The assumption appears to be quite reasonable based on the small number of countries where this information is complete. 48  The component on primary education is coded zero if the answer is no, one if the answer is yes but only for recognized individuals, and two if yes for all individuals. The component on secondary education is coded zero if the answer is no, one if access to secondary education is guaranteed, and two if access to secondary and post-secondary education is guaranteed. 40 Figure 3.26 Schooling before the Pandemic vs Learning during the Pandemic among Refugees, 3.27 (%) by Country Figure 100 ECU JOR 80 Learning during pandemic (%) 60 MEX UGA 40 KEN CRI 20 TCD ETH 0 0 20 40 60 80 100 Schooling before pandemic (%) Source: Staff calculation using HFPS. Education policy scores use data from the DWRAP database for 2017. Note: Yellow and red marks denote the country’s scoring on refugee education policies. The policy score is an aggregate sum of two components: access to primary education and access to secondary education. Each component can take on a value of zero, one, or two, depending on the level of access provided. The score ranges from a minimum of zero to a maximum of four, with higher scores indicating better access. Countries with a score of 4 are marked in yellow and those with a score of 0 or 1 are marked in red. The 45-degree line indicates where educational engagement would be if all students enrolled before the pandemic continued learning during the pandemic. 41 Financing For CHAPTER 4 Displaced Populations An 18 year-old from Daraa, Tawjihi (end of high school) 42 students in Zaatari Camp, Jordan. © UNHCR/Shawkat Alharfoush, August 2023 D isplaced populations create significant social, economic, and political pressures on the host countries. However, not only are the obligations of hosting displaced populations very unequally shared across countries (as shown The size of fiscal spending in response to the pandemic was very low among both the HFPS countries studied in this report as well as the broader set of countries that bear a large hosting burden. Except for Chad, COVID-related spending in Section 1), but hosting countries also respond in as a share of GDP in the 14 HFPS countries was different ways as seen in the wide heterogeneity even lower than the LIC average, which was already in forced displacement policies adopted (Blair, lower than that in MICs. Iraq recorded the lowest Grossman, and Weinstein 2021). This section turns spending of all HFPS countries at just 0.2 percent attention to the importance of sustainable financing of GDP, while Chad recorded the highest spending to support crisis response and integration49 for FDPs. at 5.3 percent. More spending was allocated for As seen below, there was a decline in aid intended non-health purposes, although the health response for displaced populations especially in the first to the pandemic often commanded a significant year of the pandemic, which is consistent with the budget share (Figure 4.1). unmitigated welfare impact observed in Section 3. Access to external financing was an important The role of fiscal policies received considerable determinant of fiscal spending during the attention during the pandemic. The widespread pandemic, most of which came from multilateral socioeconomic impact of the pandemic prompted organizations, such as the World Bank, IMF, an expansionary fiscal policy in many countries, regional development banks, and other UN especially in 2020 (IMF 2020). However, it has also agencies in the form of concessional loans and been widely documented that the fiscal response grants (World Bank 2022). ODA accounted for a was uneven across countries (World Bank 2022). large share of government expenditure in many In the many lower-income countries that had been of these countries (Figure 4.2). It is challenging to dealing with low growth and high levels of debt identify how forced displacement situations are distress even before the crisis, the pandemic- funded between domestic and external sources of induced shock to economic growth and falling financing, but it is very likely that the dependence revenues further constrained their fiscal space. For on external aid is high for most hosting countries.50 developing countries bearing a large share of the hosting burden, these structural challenges and Analyzing ODA data from the OECD’s CRS consequent vulnerabilities significantly undermined reveals that aid for displaced populations fell in their ability to support displaced households. 2020—at the peak of the pandemic when needs 49  Recognizing the importance of the other two channels of durable solutions, safe repatriation and third-country resettlement, this report focuses on integration, as that is the channel most of the displaced encountered during the crisis. 50  These public expenditures are often not well documented. Information systems rarely record any costs incurred as displaced people cycle through the asylum/refugee system (see, for example, Uganda in UNDP 2017). Some benefits are difficult to assign a monetary value, such as land allocated to promote self-reliance. However, available data suggest that LICs spend little on refugees; for example, Kenya allocated 0.01 percent of GDP in its fiscal year 2021/2021 budget (see Kenya 2022). In comparison, Ecuador, as an upper-middle-income country, has spent 0.3 percent of GDP per year on integrating Venezuelan migrants in recent years (see Arena et al. 2022). 43 Figure4.1 Figure 4.1 The Fiscal Response during COVID-19, by Country and Country Groups (% GDP) 10.0 9.0 8.0 7.0 Share of GDP (%) 6.0 5.0 4.0 3.0 2.0 1.0 0.0 IRQ MEX ECU JOR CRI UGA BGD SOM KEN ETH DJI BFA DRC TCD LICs LMICs UMICs IDA-18 RSW/GCFF Health Nonhealth Source: Staff calculation using IMF, “Database of Fiscal Policy Responses to COVID-19,” https://www.imf.org/en/Topics/imf-and-covid19/ Fiscal-Policies-Database-in-Response-to-COVID-19. 4.2 Net ODA Received in 2021 (%) Figure4.2 Figure 70 60 Net ODA REceived (%) 50 40 30 20 10 0 Burkina Faso Congo, Dem. Rep. Chad Ethiopia Somalia Uganda Bangladesh Djibouti Kenya Costa Rica Ecuador Iraq Jordan Mexico LIC LMIC UMIC Net ODA received (% of central government expense) Net ODA received (% of GNI) Source: “World Development Indicators,” https://databank.worldbank.org/source/world-development-indicators. Note: Net ODA measures disbursement flows (net of repayment of principal) that meet the Development Assistance Committee’s (DAC) definition of ODA and are made to countries and territories on the DAC list of aid recipients. The estimates for Mexico are both less than 0.5 percent. Net ODA received (percent of central government expense) is missing for DRC, Chad, Somalia, and Djibouti. were high. Aid to displaced populations is derived that are being tracked under the UNHRC Global using annual CRS disbursement-level data and Compact on Refugees (GCR) rely on different data identifying development aid flows intended for sources or methods to proxy refugee financing.52 displaced populations. The primary strategy entails Estimates using this strategy suggest that between keyword extraction based on project identifiers 2019 and 2020, aid flows allocated to displaced consisting of project titles and descriptions (see populations took a downturn, falling from US$9.26 Annex 4 for details).51 For comparison, the indicators billion to US$9.12 billion globally, and from US$2.91 51  Any disbursements are counted that include certain keywords, such as refugee, displaced, FDP, returnee, migration, conflict, or UNHCR, and are therefore most likely intended for displaced populations. Estimates counting disbursements tagged to specific CRS sectors are employed as a secondary measure, although the key results are largely the same. 52  A dedicated OECD survey on financing refugee situations among members is used to monitor a subset of GCR indicators, such as “Total ODA disbursements from Development Assistance Committee (DAC) donors for the benefit of refugees (and host communities) in developing countries.” For the survey, member countries used their own methods to approximate ODA going to refugee situations; “Total ODA disbursements from DAC donors for the benefit of refugees in developed countries” is estimated with a separate sector code in the CRS, and the “Number of donors providing official development assistance (ODA) to, or for the benefit of, refugees and host communities in refugee-hosting countries” uses OECD DAC Statistics on Resource Flows to Developing Countries. For details, see UNHCR (2019b) and Hesemann. Desai, and Rockenfeller (2021). 44 billion to US$2.75 billion in IDA-18/19 and Global 4.4 shows that overall aid flows had been on an Concessional Financing Facility (GCFF) countries upward trend between 2016 and 2019 and then combined. The most notable downturn was received a significant boost in 2020. Global real experienced by major hosting countries—those aid disbursements rose from US$219 billion in 2019 that host more than half a million FDPs—where real to more than US$270 billion in 2020, an increase disbursements fell from US$6.6 billion to US$6.0 of about 22 percent. The increase in the IDA-18 billion.53 There is a marked gap between the Regional Sub-Window (RSW) and GCFF countries actual aid flows for displaced populations and the and in major hosting countries was of similar expected amounts extrapolated from linear trends magnitude, where real disbursements increased based on the years leading up to the pandemic. from US$43 billion to about US$53 billion, and Figure 4.3 presents these trends globally (panel from US$92 billion to US$111 billion, respectively. A), among World Bank IDA-18 and GCFF countries Consequently, the aid effort in 2020 and 2021 (panel B), major hosting countries (panel C), and was noticeably above the linear time trend from the 14 HFPS countries (panel D). The list of the top years leading up to the pandemic. This increase in ten recipient countries between 2016 and 2021 is total aid is consistent with trends during previous shown in Table A4.4. crises when aid would first rise to finance urgent needs and then fall two–three years after the crisis. This drop in aid to displaced populations occurred ODA had thus long been a dependable and much- even as overall aid to developing countries needed source of financing during crisis times increased globally by over 20 percent.54 Figure (Ahmad and Carey 2021). Figure 4.3 Trend in Aid for Displaced Situations in Recent Years (constant 2020 US$ million) Figure 4.3 Panel A: GLOBAL Panel B: IDA−18 RSW & GCFF Panel C: IBRD/IDA HOSTS Panel D: HFPS COUNTRIES 12,000 4,000 8,000 4,000 US$ million (2020 constant) 7,000 10,000 3,000 3,000 6,000 8,000 2,000 2,000 5,000 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021 Linear trend (2016−2019) Predicted 2020/21 Actual 2020/21 Expectation gap Source: Staff calculations using OECD CRS disbursement data. 53  The IDA-18 Regional Sub-Window (RSW) represents the IDA’s additional dedicated funding to help 14 LICs that host a large number of refugees. The countries included are Bangladesh, Burkina Faso, Burundi, Cameroon, Chad, DRC, Republic of Congo, Djibouti, Ethiopia, Mauritania, Niger, Pakistan, Rwanda, and Uganda. Countries included in the IDA-18 RSW are also eligible for the IDA-19 Window for Host Communities and Refugees (WHR) (see World Bank, “IDA18 Regional Sub-Window for Refugees and Host Communities,” https://ida.worldbank.org/en/replenishments/ida18-replenishment/ida18-regional-sub-window-for-refugees-host-communities). The GCFF is a World Bank fund to support programs targeting displaced populations in Colombia, Ecuador, Jordan, Lebanon, and Moldova (GCFF 2021). In 2020, IDA-18 RSW and GCFF countries combined hosted 38.5 percent of all refugees, IDPs, and asylum seekers. The International Bank for Reconstruction and Development (IBRD)/IDA hosts refer to IBRD/IDA member states that were home to more than half a million FDPs as of 2022 (based on UNHCR refugee population statistics). 54  Aid is estimated from OECD microdata by totaling disbursements made by 30 OECD DAC member countries, 65 multilateral organizations, and 25 non-DAC countries (see OECD, “Development Assistance Committee (DAC),” https://www.oecd.org/dac/ development-assistance-committee/). Further, disbursements made by 39 large private donors are included to better represent the total aid dependance of recipient countries. 45 Figure 4.4 Trend in Total Aid Flows in Recent Years (constant 2020 US$ million) Figure 4.4 Panel A: GLOBAL Panel B: IDA−18 RSW & GCFF Panel C: IBRD/IDA HOSTS Panel D: HFPS COUNTRIES 300,000 60,000 120,000 50,000 50,000 100,000 250,000 40,000 40,000 200,000 80,000 30,000 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021 Linear trend (2016−2019) Predicted 2020/21 Actual 2020/21 Expectation gap Source: Staff calculations using OECD CRS disbursement data. Figure 4.5 Aid to displaced populations, as proportion of total aid, by region (%) Figure 4.5 10 Share of total aid (%) Global EAP ECA 5 LAC MENA SAR SSA 0 2016 2017 2018 2019 2020 2021 Source: Staff calculations using OECD CRS disbursement data. Note: East Asia and Pacific (EAP), Europe and Central Asia (ECA), Latin America and the Caribbean (LAC), Middle East and North Africa (MENA), South Asia (SAR), Sub-Saharan Africa (SSA). Aid to displacement situations measured as a Unpredictability of the source, amount, or timing proportion of total aid also fell across all regions of financing undermines the ability to plan any from 2019 to 2020 (Figure 4.5). Average real aid longer-term, sustainable solutions for this highly per displaced person declined in 2020, globally by vulnerable group. Although there was some year- about 9 percent from US$129.3 to US$118.0, and in to-year variation in total aid, the amount received all regions except Latin America and the Caribbean in 2020, which includes some of the most acute (Figure 4.6). These trends reversed in some regions phases of the pandemic, fell significantly below in 2021, but not in all. expected levels extrapolated from prepandemic trends. Strikingly, even as overall aid efforts These trends point to a key challenges fomented by increased, FDPs were disproportionately neglected ad hoc and unpredictable financial arrangements. as disbursements fell for the first time since 2016, 46 even though they were highly affected by the instrument to help establish reliable financing for pandemic. The World Bank’s IDA-18/19 windows situations of internal displacement—despite there for host and refugees and the GCFF can be used being roughly three times more displaced people to partially mitigate gaps in development spending who have not (yet) crossed an international border for refugee situations; however, there is no similar as those who have. 4.6. Aid to displaced populations, per displaced person, by region (constant 2020 US$) Figure 4.6 300 US$ (2020 constant) 200 Global EAP ECA LAC MENA SAR 100 SSA 0 2016 2017 2018 2019 2020 2021 Source: Staff calculations using OECD CRS disbursement data. Note: East Asia and Pacific (EAP), Europe and Central Asia (ECA), Latin America and the Caribbean (LAC), Middle East and North Africa (MENA), South Asia (SAR), Sub-Saharan Africa (SSA). 47 Discussion and CHAPTER 5 Conclusion Ali Mayhoub and his family of six members including 48 four children fled Yemen in 2015, Djibouti. © UNHCR/ Jordi Matas, November 2018  T his report uses newly harmonized contemporaneous HFPS data on host and displaced populations from 14 countries to provide insights into the differential welfare impact on these groups during the recent COVID much more widespread than indicated by outright employment losses alone. Welfare losses went largely unmitigated as external support fell short of needs. FDPs were crisis. Although the COVID-19 pandemic prompted often more likely than hosts to receive social a global crisis of historical scale, it also created an assistance during the pandemic. However, the opportunity for a large-scale data collection effort on data do not specify the magnitude of that social displaced populations that are often excluded from assistance or whether it was from existing programs such efforts and for whom little systematic data exist. or new initiatives specifically designed to mitigate The harmonized cross-country database used in this the impacts of the compounding crises. The reliance report combines a number of data collection efforts on negative coping strategies suggests that at the country level, making them more comparable. mitigation was not sufficient; although estimates The database covers 14 countries from different varied across countries, there were significant regions, populations of concern (IDPs, refugees, numbers of households that were forced to reduce hosts), and accommodation types (in camps, out of consumption, draw down emergency savings, or camps), allowing for a more comprehensive analysis sell their assets during the pandemic. An increase across countries and subgroups. The 14 countries in in food insecurity and malnutrition was reported this study collectively host roughly a quarter of the globally as key agricultural commodity prices more global displaced population. than doubled between 2020 and 2022. These trends are underscored in the HFPS data, which Although FDPs were deeply affected by the show that well over half of displaced households shocks that rippled through the global economy, ran out of food in 2020 and 2021, and in the most they often started out from a worse baseline and disturbing outcome, about a third of households were impacted by other contextual factors that reported members that went a full day or more contributed to the worsening of their welfare without eating because of a lack of resources. during the pandemic. Compared to a number of earlier studies that documented the pandemic’s The lack of support during the pandemic—and the impact on displaced populations, the analysis in challenges with supporting FDPs in general—is this report has been deepened on several fronts related to the fact that the vast majority of FDPs to provide a more holistic view of how welfare are hosted by LIMCs, many of which struggle with evolved during that time, for both hosts and their own development challenges. Displaced displaced populations. The key findings show populations create significant social, economic, that FDPs typically experienced larger initial and political pressures on the host countries, many employment losses that were followed by a slower of which were grappling with their own structural recovery. In addition, there were significant job economic challenges even before the pandemic, changes among those who remained employed, including fragility, low economic growth, and high again with greater turnover among FDPs. levels of poverty. These countries also tend to have Household income dynamics, where available, a high reliance on external financing for government further corroborate that the welfare impact was expenditure, including on displacement financing. 49 Figure 5.1 Poverty Figure 5.1 Trends by Country Groups, 2015–30 (%) 45 40 35 30 Poverty rate (%) LIC 25 LMIC FCS 20 Global 15 10 5 0 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 World Bank, “Poverty and Inequality Platform,” https://pip.worldbank.org/home; Mahler, Yonzan, and Lakner r al., (2022); and staff estimates. Note: FCS = fragile and conflict-affected situations. All country classifications are as of FY2023. The estimated amount of aid intended for their extreme poverty levels at around 30 percent. displaced populations declined in 2020, at the Projections suggest limited improvements in height of the pandemic, even as overall ODA poverty rates through 2030 (Figure 5.1), which are financing increased, which may explain why likely underestimated given the exclusion of FDPs in needs went largely unmet. Novel analysis using most household surveys in major hosting countries. disbursement-level data from the OECD’s CRS Recent work by Corral et al. (2020) suggests that shows that aid flows attributable to displaced accounting for displaced populations missing from populations took a downturn in 2020 during the surveys could add 30 million more people to the most acute phases of the pandemic. This coincides global count of extreme poor. with funding shortfalls that were widely and frequently reported by agencies such as UNHCR But it does not have to be this way. Better policies to at the frontline of delivering assistance to the integrate displaced populations into host societies displaced. Although overall aid levels, including and to share the hosting burden more equally can those intended for displaced populations, increased reorient these trends to a very different trajectory. in 2021, the results from this exercise highlight a key This is also consistent with the messages in the challenge of displacement financing, which is that latest World Development Report 2023: Migrants, it is highly unpredictable and may not be available Refugees, and Societies. when needed the most—as was the case during the early stages of the pandemic. Most of the countries Access to economic opportunities will allow the analyzed in this report spent less than the average displaced to become more self-reliant, which LIC to deal with the epidemiological and economic should also reduce the burden of hosting. Analysis consequences of the pandemic. in this report suggests that countries with policies that allowed access to labor markets tended to have These compounding crises had significant and higher employment levels among the displaced. potentially long-lasting impacts on the displaced Indeed, a review of existing research suggests and their hosts, which could lead to higher that displacement can benefit host communities, poverty and inequality for a generation. The latest with relatively little evidence of negative labor trends suggest that poverty has been stagnant market impacts on hosts due to the influx of FDPs.55 or on the rise in countries officially classified as Improving their legal status and providing access fragile and conflict-affected situations, putting to economic opportunities are critical elements in 55  See B. Gillsäter, ”People Feeling Conflict Don’t Want Aid—They Want Work,” Joint Center on Forced Displacement, January 19, 2023, https://www.jointdatacenter.org/people-fleeing-conflict-dont-want-aid-they-want-work/. 50 their recovery from the recent crises and in their long-term impacts, as the physiological and cognitive long-term well-being.56 Integration can also help effects of acute malnutrition, including impaired improve widespread negative perceptions about neurodevelopment and academic achievement, refugees (Alan et al. 2021). As repatriation is looking can last decades.57 However, children who benefit increasingly less likely, burden sharing through from timely catch-up growth may be insulated from resettlement or integration is critically needed. negative outcomes, underscoring the importance of monitoring efforts and prompt interventions (see, Another key aspect to the integration and for example, Martins et al. 2011). promotion of self-reliance is allowing refugee children access to national education systems Sustainable financing solutions that allow for and relieving the many social and economic continued investments and longer-term planning constraints to their learning. This is an important will be critical to easing the burden on major hosting policy area because displaced populations are communities. Many hosting countries rely on ODA disproportionately children: less than a third of the for government spending, and though detailed global population are children but among refugees, data are difficult to come by, it is very likely that it the share exceeds 40 percent. Similarly, about 20 is also a major source of funding for displacement out of a total of 55 million IDPs are children under situations. It is therefore particularly problematic the age of 15 (IDMC 2021). While remedial support is that funding to support some of the poorest and needed by all following the crisis, displaced children most vulnerable populations decreased during a are in an even more disadvantaged position due to crisis that was unprecedented in recent decades. the lack of financial stability and their heightened Financing arrangements need to be predictable vulnerability (IOM 2020). and reliable for planning purposes beyond the short term (Schady et al. 2023). Financing gaps are not Permanent losses in human capital among likely to diminish in the near future, given record- this generation could diminish future earnings high displacement levels and the increasingly potential and reduce economic mobility. Where protracted nature of displacement. Shifting the data do exist, there are signs that the pandemic balance of support more toward development aid may have led to longer-term erosions in human and adopting more inclusive refugee policies could capital. The return to school was slow in some help ease the overall burden. countries such as Ethiopia, where multiple rounds of data were collected. Data from Jordan showed Finally, the complex nature of the challenges that pandemic-related school closures led to faced by displacement situations requires more deteriorations in early grade literacy and numeracy and better data that can be relied upon to design among vulnerable children, including refugees better policies. This injunction is echoed in recent (UNICEF 2022). However, not much is known about global agreements and instructions including the the learning recovery more generally. Although Global Compact on Refugees, the establishment studies specifically on displaced populations have of the World Bank – UNHCR Joint Data Center on not yet been carried out, a few global studies on the Forced Displacement,58 a recent report on Informing potential implications of the education crisis have Durable Solutions for Internal Displacement by the led to dire predictions. The sudden and massive World Bank (World Bank 2019a), and the World education shock during the pandemic could lead Development Report 2023 on Migrants, Refugees students in LIMCs to lose up to 10 percent of and Societies (World Bank 2023d). Statistical their future earnings if the widened gaps are not inclusion of displaced people is a critical element of addressed (Schady et al. 2023). The rise in food the broader inclusion agenda, despite the technical insecurity during the pandemic may have profound and budgetary challenges of including FDPs in data 56  The most recent World Development Report recognizes the tension between gaining legal status and accessing economic opportunities as the root cause of the challenges in resolving refugee situations. Legal status is usually acquired through resettlement or return to country of origin, and therefore it is typically difficult to achieve both. The report offers examples of innovative solutions that include regional freedom of movement or a shift to labor migrant status (see World Bank 2023d). For an overview of labor market policies on displaced and host populations, see Ginn (2023). 57  See, for example, Grantham-MacGregor (1995), Tanner et al. (2015), Adebisi et al. (2019), and Kirolos et al. (2022) 58  See www.jointdatacenter.org 51 collection efforts. Robust data is critical to identifying resources or capacity (or both) to include displaced vulnerabilities and addressing the humanitarian and populations in data collection efforts. As with development challenges and helping host countries working with NSOs on general data collection, FDP and humanitarian and development institutions efforts can be a way to increase NSO capacity by quantify the burden that is to be shared—as agreed mentoring staff in a “learning by doing” approach; under the GCR, for example. capacity building ideally begins early in the process to develop frames, skills, and relationships that The experience of collecting and harmonizing allow for rapid data response. Integration of FDPs data on displaced populations during the in data systems can help fill important data and pandemic provides valuable lessons for future analytical gaps and can form ready-made sampling efforts. Chiefly, high frequency phone surveys can frames for phone surveys used for monitoring. provide rigorous, representative data. Although phone survey instruments are necessarily shorter Collecting comparable, contemporaneous data on than face to face surveys and cannot control the host and displaced populations is perhaps most interview environment, they are useful in settings in easily done as a part of standard national surveys. which in-person interviews are not possible due to Such surveys should employ a sampling strategy safety concerns or are prohibitively expensive. that includes an FDP sub-sample of sufficient size to give statistical power. Questionnaires for surveys The reliability of phone survey data is predicated covering host and FDP populations should include upon minimizing sample bias. This can be the short module of refugee or IDP identification achieved by a) using a sampling frame of the questions recommended by the UN Statistical complete universe of the population under Commission’s Expert Group on Refugee, IDP and study, b) drawing a probability sample from that Statelessness Statistics (EGRISS) to accurately frame, and c) applying reweighting techniques to identify or verify displacement status of persons compensate for any sample bias observed in the in the samples (EGRISS 2023b). 60 When such data. The availability of complete sampling frames national face to face surveys are later combined has historically been a challenge when working with subsequent phone surveys, the data can tell with displaced populations, but UNHCR’s ProGres compelling stories of changes over time, as has database can be a powerful resource in contexts been done with host and refugee populations in where it is current and complete. Institutional data- Chad.61 sharing agreements between agencies, such as the recently concluded Global Data Sharing Framework Data collection should be designed to allow Agreement between the World Bank and UNHCR, comparisons between populations, over time, can facilitate access to such databases.59 and across contexts. The contemporaneous data collection on host and displaced populations in Working in conjunction with National Statistical these COVID-era HFPS exercises afforded the Offices to collect data on displaced populations rare opportunity to compare these groups and can yield important benefits. By collaborating with benchmark them with each other.62 them, NSOs—and by extension national and local governments—are more likely to accept results. The The timing and frequency of data collection experience of unsuccessful attempts to work with significantly affect the utility of the data. For countries to collect this FDP data highlighted that analysis over periods of crisis or recovery, metrics even when there was political will to collect such are critical. Ideally, baseline data would have been data, in some instances the NSO simply lacked the collected on all sizable displaced populations before 59 See https://www.unhcr.org/news/press-releases/world-bank-unhcr-data-sharing-agreement-improve-assistance-forcibly- displaced 60 See https://egrisstats.org/ for more. 61  https://www.jointdatacenter.org/refugees-in-chad/ 62  For example, harmonized, contemporaneous host/FDP data collection can be used to inform the 12 policy priority indicators recommended by the UN Statistical Commission to be disaggregated by displacement status are 1.2.1, 1.4.2, 2.1.1, 3.1.2, 4.1.1, 6.1.1, 7.1.1, 8.3.1, 8.5.2, 11.1.1, 16.1.4 and 16.9.1, covering topics of poverty; property rights; access to health care, sanitation, and electricity; employment; adequate housing; and identity.  See https://unstats.un.org/sdgs/metadata/ and https://unstats.un.org/wiki/display/ sdgGoodPractices/Agencies+and+other+groups%3A+data+disaggregation#Agenciesandothergroups:datadisaggregation-e. RefugeesandInternallyDisplacedPeople(IDPs) 52 the pandemic to allow for pre-crisis comparisons; balanced against standardization in cases where without pre-crisis baselines, interpretation of cross-country comparisons are important. Use of contemporary outcomes and recovery trajectories a standardized questionnaire can help improve can be limited. Developing rigorous (often face-to- comparability across contexts with more indicators, face) baselines also affords an early opportunity for improve timeliness of cross-country results (by NSO capacity building. Additionally, collecting data saving on harmonization) and allowing for more at regular intervals on indicators that have high automation in data cleaning. Applying the standards variation or measurement error can help uncover developed in the EGRISS Compiler’s Manual can important trends in welfare dynamics. In practice, help (EGRISS 2023a). this calls for more frequent microdata collection that in turn should also facilitate regular welfare Considerably more research is needed on this monitoring. vulnerable and growing population. The 2023 WDR calls for data on refugees that is harmonized, Finally, the insights gained from these data are longitudinal, open, and innovative in developing compounded because results can be compared new types of surveys to inform policies. The phone across contexts. The surveys were based on a survey experience during the COVID-19 pandemic loosely standardized common questionnaire, and has shown that such data collection efforts can the data were then harmonized across countries be done in a way that is not only statistically after it was collected. Yet, harmonization was a long rigorous but also time and cost efficient—and can and resource intensive process, suggesting that provide actionable insights on some of the most country customization may need to be carefully marginalized communities. 53 Annexes ‘Aleppo’, the furniture workshop founded by 54 Venezuelan whose grandparents fled Syria. Ecuador. © UNHCR/Jaime Giménez, September 2022  Annex 1. Country Surveys N.B. An * indicates rounds that were included in the analysis in this report. Dates listed are the approximate start dates of survey rounds. BANGLADESH Rounds included Hosts & FDPs Refugees in analysis in this Rounds Sample size Sample size report: 1-3 Date Date (HH) (HH) 1* Apr-20 1,816 Apr-20 1,358 Rounds and 2* Oct-20 2,180 Oct-20 1,662 Sample Size 3* Apr-21 2,194 Apr-21 1,458 The sampling frame is a longitudinal, integrated, nationally representative household survey? NO The Cox’s Bazar monitoring surveys use the Cox’s Bazar Panel Host Survey (CBPS) baseline as the sampling frame. The CBPS was a Sampling Frame face-to-face survey fielded in 2019 that used the 2011 population census and GIS data as a sampling frame for hosts and the IOM FDPs NPM12 (International Organization for Migration, Needs and Population Monitoring) Round 12 data for Rohingya refugees. Host Host population within Cox’s Bazar and Bandarban district Coverage Rohingya population living in camps within Cox’s Bazar and FDPs Bandarban district The CBPS study was divided among three strata covering Rohingya refugees in camps and host communities in Cox’s Bazar district and some adjacent regions of Bandarban district. The CBPS High- Frequency Tracking attempted to follow the full baseline sample of 5,020 household in each round, with no alterations or additions made to the sampling design. For hosts, a two-stage sampling strategy was followed. The first stage of selection was done at the mauza level by strata. A random Host sample of 66 mauzas was drawn from a frame of 286 mauzas Sampling using probability proportional to size. Based on census population Strategy and size, each mauza was divided into segments of roughly 100-150 Representativeness households. The second stage selected three segments from each chosen mauza with equal probability of selection. Within each selected power supply unit in camps (blocks) and hosts (mauza- segments), all households (100–150 on average) were listed. Of listed households, 13 households were selected at random for an interview, with an additional replacement list of 5 households. Stages of sample selection: For camps, NPM12 divided all camps into 1,954 majhee blocks. 1,200 blocks were randomly selected using FDPs a probability proportional to the size of the camp. A full listing was carried out in each selected camp block. Modules Access to Basic Needs, Labor, Education URLs https://microdata.worldbank.org/index.php/catalog/4528 55 BURKINA FASO Rounds included Hosts IDPs in analysis in this Rounds Sample size Sample size Date Date report: 10-12 (HH) (HH) 1 Jun-20 1,968 2 Jul-20 2,037 3 Sep-20 2,013 4 Nov-20 2,011 5 Dec-20 1,944 6 Jan-21 1,985 7 Feb-21 1,979 8 Mar-21 1,967 Rounds and 9 Apr-21 1,971 Sample Size 10* May-21 1,998 May-21 1,146 11* Jun-21 1,986 May-21 1,107 12* Apr-22 1,971 Jun-21 1,043 13 Jun-22 1,735 14 Aug-22 1,708 15 Oct-22 1,700 16 Dec-22 1,688 17 Mar-23 1,642 The sampling frame is a longitudinal, integrated, nationally representative household survey? YES 2018/19 EHCVM Host Enquete Harmonisée sur les Conditions de Vie des Ménages CONASUR Database Sampling Frame The CONASUR database (developed and supported by the government of Burkina Faso with the technical and financial support of FDPs development partners, including UNHCR, IOM and OCHA) is updated regularly and has an exhaustive list of refugees and IDPs, along with a few socio-demographic characteristics, as well as information on the phone numbers of households. Host National Coverage FDPs IDPs (in 9 regions out of 13) Households from the 2018/19 EHCVM with at least one valid phone number established the sampling frame for the high-frequency survey (HFS). To obtain representative strata at the national, capital (Ouagadougou), urban, and rural levels, the target sample size for the Host HFS was 1,800 households (assuming a 50% non-response rate, the minimum required sample is 1,479). To account for non-response and attrition, 2,500 households were called in the baseline round of the HFS. 1,968 households were fully interviewed during the first round of Sampling interviews. Strategy and The BFA HFPS-IDPs was representative of households that have Representativeness access to phones. Taking that into consideration, a key concern was the bias introduced by sampling households with at least a phone number, as phone penetration in some regions/areas might be limited. However, according to data from the CONASUR database, the FDPs percentage of households with at least one phone number was very high, accounting for above the 74% in all the sampled regions. To account for non-response and attrition, 1,500 households were selected in the baseline round of the HFS. 1,166 households were fully interviewed during the first round of interviews. Household Roster, Knowledge Regarding the Spread of COVID-19, Behavior and Social Distancing, Access to Basic Needs, Education, Credit, COVID Testing and Vaccination, Employment and Income, Food Security, Shocks, Fragility, Conflict Modules and Violence, Other revenues, Social protection, Personal Health Questionnaire, Displacement, Early Child Development - Parental Support, Concerns, Economic Sentiment, Price of Items, Climate Change https://microdata.worldbank.org/index.php/catalog/3768 URLs https://microdata.worldbank.org/index.php/catalog/4481 56 CHAD Rounds included in Hosts Refugees the analysis in this Rounds Sample size Sample size report: 3-4 Date Date (HH) (HH) 1 May 202 1,748 Rounds and 2 Jul-20 1,708 Sample Size 3* Jan-21 1,609 Jan-21 919 4* Mar-21 1,482 Mar-21 852 The sampling frame is a longitudinal, integrated, nationally representative household survey? YES (for N and R) 2018/2019 ECOSIT 4 Host Enquête sur la Consommation des Ménages et le Secteur Informel Sampling Frame au Tchad 2018/2019 RHCH FDPs Refugees and Host Communities Household Survey in Chad (subsample of ECOSIT 4) Host National Coverage FDPs Refugees The sampling of the high-frequency survey aimed at having representative estimates nationally and by area of residence: Host Ndjamena (capital city), other urban and rural areas. The minimum sample size was 2,000, out of which 1,748 households (87.5%) were Sampling successfully interviewed at the national level. Strategy and Representativeness ECOSIT 4 contained a subsample of Chadians and refugee households from which the refugee sample of this high frequency FDPs survey was drawn. Sampling weights were adjusted to ensure that the two samples were representative of all Chadian households and all refugee households, respectively. Household Roster, Knowledge of COVID-19, Behavior and Social Distancing, Employment and Income, Access to Basic Services, Income Loss, Subjective Modules Poverty, Prices and Food Security, Shocks/Coping, Impacts of COVID-19, Social Safety Nets and Assistance, Perception, Impacts of COVID-1, Fragility and Security, Vaccine; Gender-Based Violence URLs https://microdata.worldbank.org/index.php/catalog/3792 57 COSTA RICA Rounds included Hosts FDPs in analysis in this Rounds Sample size Sample size report: 10-12 Date Date (HH) (HH) Phase 1 1 May-20 801 - - 2 Jul-20 636 - - Rounds and 3 Jul-20 658 - - Sample Size Phase 2 - - 4* May-21 802 Mar-21 1,163 5‡ Oct-21 905 Jul-21 761 The sampling frame is a longitudinal, integrated, nationally representative household survey? NO Sampling Frame RDD Host Random Digit Dialing protocol FDPs UNHCR ProGres Database Host National Coverage FDPs National with stratified random sampling The sample was based on a dual frame of cellphone and landline numbers generated through a Random Digit Dialing (RDD) process. The RDD methodology produces all possible phone numbers in Host the country under the national phone numbering plan and draws a Sampling random sample of numbers. This method ensures coverage of all Strategy and landline and cellphone numbers active at the time of the survey. Representativeness National representation of PoCs registered in UNHCR ProGres database, with additional stratified sampling for Nicaraguan PoCs FDPs in the Greater Metropolitan Area (GAM), Venezuelans in the GAM, Cubans in the GAM, and Nicaraguan PoCs in the North. Hosts: Basic Information, Knowledge Regarding the Spread of COVID-19, Behavior and Social Distancing, Access to Basic Services, Employment, Income Loss, Food Modules Security, Concerns, Coping Strategies, Social Safety Nets, Trust FDPs: Knowledge, Behavior, Access, Employment, Income, Food Security, Concerns, Resilience, Networks, Demographics Phase 1 https://microdata.worldbank.org/index.php/catalog/4052 Phase 2 https://microdata.worldbank.org/index.php/catalog/4562 URLs Phase 2 https://microdata.worldbank.org/index.php/catalog/4755 Report LAC https://openknowledge.worldbank.org/handle/10986/35902 UNHCR https://microdata.unhcr.org/index.php/catalog/636 Note: ‡ The Oct-21 host data were not published in time to be harmonized in this effort and so are not used in the analysis in this report. 58 DEMOCRATIC REPUBLIC OF CONGO Rounds included Hosts Refugees IDPs in analysis in this Rounds Sample Sample Sample report: 10-12 Date Date Date size (HH) size (HH) size (HH) 1 Jun-2020 1,453 - - 2 Jul-2020 1,438 - - 3 Aug-2020 1,437 - - 4 Sep-2020 1,440 - - Rounds and 5 Nov-2020 1,438 - - Sample Size 6 Feb-2021 1,443 - - 7* Oct-2021 1,252 Oct-2021 126 Oct-2021 1,087 8* Nov-2021 1,261 Nov-2021 163 Oct-2021 1,057 9 Jan-2022 1,260 Jan-2022 139 Jan-2022 1,086 The sampling frame is a longitudinal, integrated, nationally representative household survey? NO Sampling Frame Host Social registry in Eastern DRC Built up by the Social Protection and Jobs (SPJ) program and managed by Fonds Social de la RDC (FSRDC) across different sites in Eastern DRC. The FDPs social registry includes both hosts and self-declared FDPs. Host Eastern DRC Coverage FDPs Refugees, IDPs (and returnees; not used in this report) in Eastern DRC The social registry was comprised of individuals showing up to the public lotteries of the program, with those selected through the public lottery becoming beneficiaries of the SPJ-FSRDC project. The program remunerated beneficiaries U$3 per day for their participation in community Host works, which was announced prior to the public lottery. As a result, the selection mechanism ensured that only individuals from poor and vulnerable populations participated in the lotteries – those who were willing and able to carry out work for the established daily wage. The SPJ-FSRDC program collected phone numbers during public lotteries. Hence, the current panel Sampling survey by the DRC Crisis Observatory was able to select from a pool of Strategy and vulnerable and poor populations residing in Eastern DRC who showed up to Representativeness the public lottery and provided a phone number to Monitoring Automated for Real Time Analysis (MARTA). MARTA recorded a total of 68,558 respondents across Beni (including Kalunguta), Bunia, Goma, Lubero, and Komanda, FDPs 51,007 of whom provided a phone number. Displacement status was self-reported in the SPJ-FSRDC registry used as the sampling frame. The interpretation is that this sample is representative of all refugees/ IDPs/returnees who self-selected into participation of the SPJ project and thus were sufficiently poor to qualify as vulnerable FDPs (showing up for daily US$3 per day wage). Access to Food and Medical Supplies; Schooling; Employment; Income; Coping Modules Strategies; Food Security; Social Assistance; COVID-19 Welfare Perceptions; Early Childhood Development; Mental Health URLs https://crisisobservatory.org/welfare-monitoring 59 DJIBOUTI Rounds included Hosts FDPs in analysis in this Rounds Sample size Sample size report: 10-12 Date Date (HH) (HH) 1 Jul-20 1,486 Rounds and 2 Sep-20 1,457 Sample Size 3* Dec-20 1,375 Dec-20 564 4* Mar-21 1,561 Oct -21 435 The sampling frame is a longitudinal, integrated, nationally representative household survey? NO 2017 National social registry Collected by the Ministry of Social Affairs (MASS), it is an official Host database of households in Djibouti that may benefit from public Sampling Frame transfers and be particular targets of poverty alleviation efforts. 2019 Refugee survey Collected in 2019 by Institut National de la Statistique et de la FDPs Démographie (INSD) jointly with MASS, WFP, and UNHCR. The original sample of the Refugee Survey in 2019 was drawn from the refugee registration data. Host Urban Coverage FDPs Djibouti-city and 3 refugee villages The sample design was a one-stage probability sample selected from the sampling frame and stratified along two dimensions: the survey domain (three categories) and the poverty status (binary). This yielded six independent strata. Within each stratum, Host Sampling households were selected with the same ex ante probability, but Strategy and this differed across strata. Initially 1,590 households were drawn. Representativeness Given a non-response rate averaging 30 percent, a replacement sample of 750 households was selected. Among the Refugees Survey Sample, the refugee sample of the FDPs COVID-19 survey was not drawn randomly but by selecting the households that had a phone number. Household Roster, Employment, Household Income Sources, Access to Basic Modules Goods, Access to Health Care and Education, Food Insecurity, Vaccine Attitudes, Gender https://microdata.worldbank.org/index.php/catalog/4216 URLs https://microdata.worldbank.org/index.php/catalog/4070 60 ECUADOR Rounds included Hosts FDPs in analysis in this Rounds Sample size Sample size report: 10-12 Date Date (HH) (HH) Phase 1 1* May-20 958 May-20 269 2* Jun-20 785 Jun-20 240 3* Jul-20 646 Jul-20 207 Rounds and 4* Aug-20 740 Aug-20 231 Sample Size Phase 2 5* May-21 951 May-21 401 6‡ Oct-21 1,032 Oct-21 583 7 Feb-22 1,072 Feb-22 445 8 Jun-22 1,106 Jun-22 356 The sampling frame is a longitudinal, integrated, nationally representative household survey? NO RDD Host Sampling Frame Random Digit Dialing protocol List of cell phone numbers FDPs Phone numbers that had contact with Venezuela (incoming or outbound) that were confirmed to be from Venezuela Host National Coverage FDPs Venezuelan households living in Ecuador The sample was based on a dual frame of cellphone and landline numbers generated through an RDD process. The RDD methodology produces all possible phone numbers in the country Host under the national phone numbering plan and draws a random sample of numbers. This method ensures coverage of all landline and cellphone numbers active at the time of the survey. Considering Venezuelans are a small part of the population in Sampling Ecuador, the strategy to identify and sample Venezuelan migrants Strategy and was different from that used for the overall population. To create a Representativeness sampling frame, a list of all cell phone numbers of customers who registered regular incoming or outgoing calls from Venezuela was FDPs generated. A first-phase simple random sample was selected from this frame and contacted to confirm that the owners were indeed Venezuelan and determine if they were willing to participate in the survey. From those who agreed to participate in the study and were confirmed as Venezuelan adults, a second-phase sample was selected to complete the survey. Cover Page, Basic Information, Knowledge Regarding the Spread of COVID-19, Modules Behavior and Social Distancing, Access to Basic Services, Employment, Income Loss, Food Security, Concerns, Coping Strategies, Social Safety Nets, Trust Phase 1 https://microdata.worldbank.org/index.php/catalog/4060 Phase 2 https://microdata.worldbank.org/index.php/catalog/4564 Phase 2 https://microdata.worldbank.org/index.php/catalog/4757 URLs Phase 2 https://microdata.worldbank.org/index.php/catalog/5406 FDPs https://microdata.worldbank.org/index.php/catalog/5665 Report LAC https://openknowledge.worldbank.org/handle/10986/35902 Note: ‡The Oct-21 rounds of data were not published in time to be harmonized in this effort and so are not used in the analysis in this report 61 ETHIOPIA Rounds included Hosts FDPs in analysis in this Rounds Sample size Sample size report: 10-12 Date Date (HH) (HH) 1 Apr-20 3,249 2 May-20 3,107 3 Jun-20 3,058 4 Jul-20 2,878 5 Aug-20 2,770 6* Oct-20 2,753 Sep-20 1,676 Rounds and 7* Nov-20 2,536 Oct-20 1,429 Sample Size 8 Dec-20 2,222 9 Dec-20 2,077 10 Feb-21 2,178 11 Apr-21 1,982 12 Jun-21 888 13 Oct-22 2,876 The sampling frame is a longitudinal, integrated, nationally representative household survey? YES (for nationals) 2018/19 ESS Sampling Frame Host Ethiopia Socioeconomic Survey (ESS) ARRA/UNHCR registration database FDPs Ethiopia Agency for Refugee and Returnee Affairs Host National Coverage FDPs Refugees in Addis Ababa, Sub-office Jijiga, Sub-office Shire To obtain representative strata at the national, urban, and rural levels, the target sample size for the HFPS-HH was 3,300 households: 1,300 in rural and 2,000 in urban areas. In rural areas, the survey team attempted to call all phone numbers included in the ESS, as only 1,413 households owned phones and another 771 Host households provided reference phone numbers. In urban areas, 3,213 households owned a phone and 224 households provided reference phone numbers. To account for non-response and attrition, all the 5,374 households were called in round 1 of the Sampling HFPS-HH. Strategy and Representativeness The geographic division of the UNHCR sub-office, combined with the phone penetration rate, was used to inform which stratification was best placed to yield robust representative results of refugee populations. The team considered only strata with a phone penetration higher than 30 percent in order to (i) have enough FDPs phone numbers and (ii) not introduce too high a sampling bias. The sample was drawn using a simple random sample without replacement. Expecting a high non-response rate based on experience from the HFPS-HH, the team drew a stratified sample of 3,300 refugee households for the first round. Household Roster, Knowledge Regarding the Spread of COVID-19, Behavior and Social Distancing, Access to Basic Services, Employment, Income Loss and Modules Coping, Food Security, Aid and Support/ Social Safety Nets, Agriculture, Locusts, WASH, Education and Childcaring, Credit, Migration, Return Migration, SWIFT, Youth Aspirations and Employment, Access to Health Services, Food Prices https://microdata.worldbank.org/index.php/catalog/3716 URLs https://microdata.worldbank.org/index.php/catalog/4543 62 IRAQ Rounds included Hosts IDPs FDPs Refugees in analysis in this Rounds Sample Sample Sample report: 10-12 Date Date Date size (HH) size (HH) size (HH) 1 Aug-20 1,621 2 Sep-20 1,621 3* Oct-20 1,623 Oct-20 728 Oct-20 1,602 4* Nov-20 1,629 Nov-20 746 Nov-20 1,406 Rounds and 5* Dec-20 1,614 Dec-20 717 Sample Size 6* Jan-21 1,651 Jan-21 720 7 Jun-21 1,627 8 Jul-21 1,635 9 Aug-21 1,628 The sampling frame is a longitudinal, integrated, nationally representative household survey? NO Sampling Frame Host 2018 MICS IDP – Data from all major Mobile Network Operators (MNOs), Refugees FDPs – UNHCR ProGres Host National Coverage IDPs and returnees in Kurdistan and Northern region FDPs Refugees: national coverage The data collection methodology consisted of a countrywide survey covering the 18 governorates in Iraq. The sample size was disaggregated by 18 governorates, and the survey firm applied a random sampling approach to reach participants from different governorates in order to reach the given geographical quotas. All major Host MNOs active in the country were included within the sampling frame to ensure a representative sample. The sample size was designed to detect changes in the prevalence of food insecurity (mainly people with inadequate food consumption) at governorate level as reported in the 2016 Comprehensive Food Security and Vulnerability Analysis (CFSVA) survey in Iraq. Sampling IDP Sample: Almost all the IDPs in Iraq are located in the three Strategy and governorates of Kurdistan region and five governorates of the Northern Representativeness region. Therefore, the coverage of the mobile phone survey for the IDP sample was limited to those two regions to create 4 strata: Duhok (stratum 1), Erbil and Sulaimaniya (stratum 2), Nineveh (stratum 3), and the rest of the northern region, i.e., Kirkuk, Diyala, Anber, and Salah Al- deen (stratum 4). FDPs Refugee Sample: The sample covered all governorates in Iraq and included households from Syria as well as households of different nationalities. The sample size and demographics were derived through a stratification process, which involved dividing the population into homogeneous subgroups before sampling. Hence, random sampling was employed for the study, using three levels of stratification: (1) governorate, (2) country of origin, and (3) camp and out-of-camp status (specifically for Syrian refugees). Demographic Section, Employment, Entrepreneurial/Business activities, Agricultural Activities, Food Consumption, Reduced Coping Strategy, Access to Food and Modules Market, Transfers, Health Status and Access to Health Services, Education/Distance Learning, COVID-19 Test and Vaccine, Household Expenses https://microdata.worldbank.org/index.php/catalog/4023 URLs https://microdata.worldbank.org/index.php/catalog/4076 https://microdata.unhcr.org/index.php/catalog/774/related-materials 63 JORDAN Rounds included Hosts FDPS in analysis in this Rounds Sample size Sample size report: 10-12 Date Date (HH) (HH) 1 Mar-21 1,004^ Rounds and 2* Nov-21 732 Nov-21 813 Sample Size 3 Apr-22 923 Apr-22 1,516^ 4 Jun-22 800 The sampling frame is a longitudinal, integrated, nationally representative household survey? NO National Unified Registry (NUR) bread subsidy applicants Sampling Frame Host The NUR is an administrative registry of potential beneficiaries for social assistance. FDPs UNHCR database Host National Coverage FDPs Syrian refugees living in the country The NUR is an administrative registry of potential beneficiaries for social assistance. The bread subsidy was estimated to cover around 80 percent of the Jordanian population up until 2021 when it was discontinued. The sample for the survey was drawn in 2020. Host Since the sampling frame tends to over-represent the poor, an ex Sampling post weight adjustment was applied to better reflect population Strategy and demographics in terms of gender, age of the household head, and Representativeness socioeconomic status. The sample was stratified by rural/urban location and camp/non- camp location in four bins: Amman, other governorates-urban, FDPs other governorates-rural, and camps. An ex post weight adjustment was also applied to the refugee population to better reflect this population’s demographics using the UNHCR database. Including: Employment, Food Security, Coping Strategies Used by Households, Modules Mental Health URLs Forthcoming Note: ^ Interview modes: Phone and face-to-face; otherwise, phone only. 64 KENYA Rounds included Hosts FDPs in analysis in this Rounds report: 10-12 Date Sample size (HH) Date Sample size (HH) 1* May-20 4,060 May-20 1,159 2* Jul-20 4,489 Jul-20 1,540 3* Sep-20 4,979 Sep-20 1,336 Rounds and 4* Jan-21 4,890 Jan-21 1,245 Sample Size 5* Mar-21 5,857 Mar-21 1,405 6* Jul-21 5,764 Jul-21 1,258 7* Nov-21 5,633 Nov-21 1,137 8 May-22 4,550 May-22 1,355 The sampling frame is a longitudinal, integrated, nationally representative household survey? YES, partly (for nationals and FDPs) 2015/16 KIHBS and RDD Sampling Frame Host Kenya Integrated Household Budget Survey SES, UNHCR database FDPs Socio Economic survey Host National Coverage Refugees and stateless: Urban refugees, Shona stateless and camps FDPs (Kakuma, Kalobeyei, Dadaab) The COVID-19 RRPS with Kenyan households had two samples. The first sample consisted of households that were part of the 2015/16 KIHBS CAPI pilot and provided a phone number. The 2015/16 KIHBS CAPI pilot was representative at the national level, stratified by county and place of residence (urban and rural areas). At least one valid phone number was obtained for 9,007 households and all of them were included in the COVID-19 RRPS sample. The second sample consisted of households Host selected using the RDD method. A list of random mobile phone numbers was created using a random number generator from the 2020 Numbering Frame produced by the Kenya Communications Authority. The initial sampling frame therefore consisted of 92,999,970 randomly ordered phone numbers assigned to three networks: Safaricom, Airtel, and Telkom. An introductory text message was sent to 5,000 randomly selected numbers to determine if numbers were in operation. Out of these, 4,075 were found to be active and formed the final sampling frame. The third RRPS sample consisted of urban and camp-based refugees as Sampling well as stateless people registered by the UNHCR. The sample aimed Strategy and to be representative of the refugee and stateless populations in Kenya. It comprised five strata: Kakuma refugee camp, Kalobeyei settlement, Representativeness Dadaab refugee camp, urban refugees, and Shona stateless, where sampling approaches differ across strata. For refugees in Kakuma and Kalobeyei, as well as for stateless people, recently conducted socioeconomic surveys (SES)were used as sampling frames. For the refugee population living in urban areas and the Dadaab camp, no such household survey data existed, and sampling frames were based on FDPs UNHCR’s registration records (ProGres), which include phone numbers. For Kakuma, Kalobeyei, Dadaab, and urban refugees, a two-step sampling process was used. First, 1,000 individuals from each stratum were selected from the corresponding sampling frames. Each of these individuals received a text message to confirm that the registered phone was still active. In the second stage, implicitly stratifying by sex and age, the verified phone number lists were used to select the sample. For the stateless population, all the participants of the Shona SES (n=400) were included in the RRPS, because of limited sample size. The sampling frames for the refugee and Shona stateless communities are thus representative of households with active phone numbers registered with UNHCR. Household Roster Background and Information, Travel Patterns and Interactions, Modules Employment, Food Security, Income Loss, Transfers, Subjective Welfare, Health, COVID Knowledge, Household and Social Relations https://microdata.worldbank.org/index.php/catalog/3774 URLs https://microdata.unhcr.org/index.php/catalog/296/ 65 MEXICO Rounds included Hosts FDPs in analysis in this Rounds Sample size Sample size report: 10-12 Date Date (HH) (HH) Rounds and 1* Feb-21 1,142 Feb-21 1,220 Sample Size 2* Aug-21 517 Aug-21 701 The sampling frame is a longitudinal, integrated, nationally representative household survey? NO Sampling Frame RDD Host Random Digit Dialing protocol FDPs UNHCR ProGres registry, stratified Comparable subsample of the national population in the same Host locations where PoCs were sampled (see below). Four strata comprising areas where PoCs are most likely to settle: 1. Coverage Southern Mexico – Honduran and El Salvadoran PoC population 2. Mexico City – Honduran, El Salvadoran, and Cuban PoC population FDPs 3. Northern and Central Industrial Corridor – Hondurans and El Salvadoran PoC population 4. Venezuelan population – Mexico City, Monterey (Nuevo Leon), and Cancun (Quintana Roo). The sample was based on a dual frame of cellphone and landline numbers generated through an RDD process. The RDD methodology produces all possible phone numbers in the country under the national phone numbering plan and draws a random Host Sampling sample of numbers. This method ensures coverage of all landline Strategy and and cellphone numbers active at the time of the survey. RDD Representativeness was used to generate a comparable subsample of the national population in the same locations where PoC were sampled. ProGres database with representative samples of the four strata FDPs of PoCs described above. The population of the four groups represents 67% of the active registered refugees in Mexico. Cover Page, Basic Information, Knowledge Regarding the Spread of COVID-19, Modules Behavior and Social Distancing, Access to Basic Services, Employment, Income Loss, Food Security, Concerns, Coping Strategies, Social Safety Nets, Trust Phase 1 https://microdata.worldbank.org/index.php/catalog/4056 Phase 2 https://microdata.worldbank.org/index.php/catalog/4568 URLs Phase 2 https://microdata.worldbank.org/index.php/catalog/4761 Report LAC https://openknowledge.worldbank.org/handle/10986/35902 66 SOMALIA Rounds included Hosts & FDPs FDPs in analysis in this Rounds Sample size Sample size report: 10-12 Date Date (HH) (HH) Rounds and 1 Jun-20 2,063 Jun-20 718 Sample Size 2 Jan-21 1,344 Jan-21 350 The sampling frame is a longitudinal, integrated, nationally representative household survey? NO Sampling Frame Host RDD FDPs Random Digit Dialing protocol Host Coverage National coverage, including nomads and IDPs FDPs The SHFPS sampled 2,811 households across Somalia using phone numbers selected through an RDD protocol. A sample allocation for Host the COVID-19 SHFPS was developed to provide representative and reliable estimates nationally, at the level of Jubaland, South West, Sampling HirShabelle, Galmudug, Puntland, Somaliland, and Banadir Regional Strategy and Administration, and by population type (i.e., urban, rural, nomads, Representativeness and IDP populations). FDPs Reaching rural and nomadic lifestyle respondents proved to be challenging and additional measures were employed to sample within that population stratum (see microdata library webpage). Household Roster, Knowledge Regarding the Spread of COVID-19, Behavior and Social Distancing, Concerns Related to the COVID-19 Pandemic, Access to Basic Modules Goods and Services, Employment, Income Loss, Remittances, Shocks and Coping Mechanisms, Food Insecurity, Social Assistance and Safety Nets, COVID-19 Vaccine, Mortality, Interaction with Internally Displaced Persons URLs https://microdata.worldbank.org/index.php/catalog/4077 67 UGANDA Rounds included Hosts FDPs in analysis in this Rounds Sample size report: 10-12 Date Date Sample size (HH) (HH) 1 Jun-20 2,227 2 Jul-20 2,199 3 Sep-20 2,179 4* Oct-20 2,135 Oct-20 2,010 5* Feb-21 2,122 Dec-20 1,852 Rounds and 6 Mar-21 2,100 Feb-21 1,985 Sample Size 7 Sep-21 1,950 8 Jun-22 1,881 9 Aug-22 1,871 10 Oct-22 1,668 11 Dec-22 1,666 The sampling frame is a longitudinal, integrated, nationally representative household survey? YES (for nationals) Sampling Frame 2019/20 UNPS Host Uganda National Panel Survey FDPs 2018 UBOS (Uganda Bureau of Statistics) survey & UNHCR database Host National Coverage FDPs Refugees in Kampala, South-West, and West-Nile To obtain a nationally representative sample for the COVID-19 Impact Survey, a sample size of approximately 1,800 successfully interviewed households was targeted. However, to reach that target, a larger pool of households needed to be selected from the frame due to non- Host contact and non-response common for telephone surveys. Thus, all the households in the 2019/20 round of the UNPS that had phone numbers for at least one household member, or a reference individual, were included in the initial sample. This consisted of 2,227 households, that is, 72% of the UNPS 2019/20 sample. Sampling The Profile Global Registration System (ProGres) served as a sample Strategy and frame for the URHFPS. It was complemented by the data collected Representativeness for the refugee household survey carried out by UBOS and the World Bank in 2018. The sample was selected from the pool of refugees with phone numbers. The targeted sample included 2,100 observations: 300 observations in each stratum. Four countries of origin were FDPs targeted in the survey: Burundi, Democratic Republic of Congo (DRC), Somalia, and South Sudan. The combination of country of origin and region were used to create seven strata: Kampala-Somalia, Kampala- other (Burundi, DRC, South Sudan), South West-Burundi, South West- DRC, South West-South Sudan, South West-Somalia, and West Nile- South Sudan. Access to Basic Goods and Services, Access to Education, Access to Health Services, Access to Medicine and Treatment, Access to Soap and Water, Agriculture, Anti-COVID-19 Behavior and Social Distancing, Assets - Climate Change Impact, Concerns Re: COVID-19 Impacts, Consumption Price of Staple Food, COVID-19 Modules Symptoms, Credit, Early Childhood Development, Economic Sentiment, Employment, Food Security, Household Composition, Income Losses, Knowledge and False Beliefs Re: COVID-19, Mental Health, Non-Farm Enterprises, Perceptions Re: Efficacy of Government Actions, Safety Nets, Shocks and Coping Strategies, Survey of Well- Being via Instant and Frequent Tracking, Vaccination and Willing to Test https://microdata.worldbank.org/index.php/catalog/3765 URLs https://openknowledge.worldbank.org/handle/10986/35819 68 FDPs FDPsas share as share national ofof population national (%) population (%) Millions Millions 0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Colombia Colombia West WestBank Bankand andGaza Gaza Syrian SyrianArab ArabRep. Rep. Figure A2.1a Figure A2.1a Syrian SyrianArab ArabRep. Rep. Congo, Congo,Dem. Dem.Rep. Rep. Jordan Jordan Türkiye Türkiye Lebanon Lebanon Yemen Yemen Somalia Somalia Sudan Sudan South SouthSudan Sudan Afghanistan Afghanistan Colombia Colombia Jordan Jordan Yemen Yemen Somalia Somalia Central CentralAfrican AfricanRep. Rep. Nigeria Nigeria Georgia Georgia Refugees Refugees West WestBank Bankand andGaza Gaza Afghanistan Afghanistan Ethiopia Ethiopia Libya Libya Iraq Iraq Sudan Sudan South SouthSudan Sudan Congo, Congo,Dem. Dem.Rep. Rep. Azerbaijan Pakistan Pakistan Azerbaijan Asylum-seekers Asylum-seekers Türkiye Uganda Uganda Türkiye Figure A2.1a Thirty LMICs Hosting the Most FDPs in 2019 Cameroon Cameroon Lebanon Lebanon IDPs IDPs Iraq Iraq Cameroon Cameroon Chad Chad Bangladesh Bangladesh FDPs (Refugees, Asylum-Seekers, IDPs, VDAs) FDPs (Refugees, Asylum-Seekers, IDPs, VDAs) Iran Iran VDAs VDAs Uganda Uganda Congo, Congo,Rep. Rep. Peru Peru Bosnia Bosniaand andHerzegovina Herzegovina Ukraine Ukraine Ecuador Ecuador India India Source: Staff illustration using UNHCR, “Refugee Data Finder,” https://www.unhcr.org/refugee-statistics/. Djibouti Djibouti Kenya Kenya Peru Peru Chad Chad Burkina BurkinaFaso Faso Central CentralAfrican AfricanRep. Rep. Guyana Guyana Burkina BurkinaFaso Faso Figure A2.1b Thirty LMICs Hosting the Most FDPs as a Share of National Population in 2019 Honduras Honduras Ecuador Ecuador Costa CostaRica Rica Libya Libya Annex 2. Supplemental Figures and Tables Mauritania Mauritania Countries with data included in this report Myanmar Myanmar Countries with data included in this report Countries with data included in this report Countries with data included in this report 69 Table A2.1 Core Modules for the COVID-19 HFPS Section/Module Description Roster of individuals living in the household; age; sex; and relationship to the Household Roster household head. This section includes questions on: (i) Respondent household’s ability to buy medicines and selected staple items that were needed in the week preceding the Access to Basic survey, and if not able, why they could not be purchased; (ii) School attendance Needs and Services status for households with school-age children; availability and use of learning activities during the school closures; (iii) access and utilization of health care services; and (iv) access to financial services. Respondent’s knowledge about the pandemic, including questions on knowledge Knowledge of ways to reduce the risk of contracting coronavirus and knowledge of steps that (COVID-19) the government has taken to reduce spread of corona virus. Behavior Selected questions on the respondent’s practices, including frequent hand (COVID-19) washing and avoiding handshake/physical greetings, avoiding gatherings. Assistance that anyone in the household received from institutions by type of Aid and Assistance assistance, amount received, and types of institutions providing the assistance. Respondent’s work status in the week preceding the survey; job loss and its Employment reasons; employers and their sectors; changes in work arrangements; profile of household-owned business and changes. Household-level questions on food insecurity experience by an adult household member for the 30 days preceding the survey. In several cases, the Food Security module adapted from the Food and Agriculture Organization’s (FAO) Food Insecurity Experience Scale was supplemented or substituted with WFP’s Food Consumption Score. Types of household income sources: farming, personal income from wage employment or pension, own non-farm business; remittances from within the Income Loss country and abroad; income from properties, investments and savings; support from government and NGOs and other charitable organizations; and changes in income sources after the outbreak. Shocks and Coping Shocks that affected households and their coping strategies. Mechanisms 70 A2.2 Mobility Trends and Policy Stringency in Countries with Phone Surveys FigureA2.2 Figure Bangladesh Burkina Faso Chad 100 100 (% change in visitors) Oxford Stringency Google Mobility 50 Index 50 0 −50 0 −100 Jan20 Jul20 Jan21 Jul21 Jan20 Jul20 Jan21 Jul21 Jan20 Jul20 Jan21 Jul21 Costa Rica Democratic Republic of Congo Djibouti 100 100 (% change in visitors) Oxford Stringency Google Mobility 50 Index 50 0 −50 0 −100 Jan20 Jul20 Jan21 Jul21 Jan20 Jul20 Jan21 Jul21 Jan20 Jul20 Jan21 Jul21 Ecuador Ethiopia Iraq 100 100 (% change in visitors) Oxford Stringency Google Mobility 50 Index 50 0 −50 0 −100 Jan20 Jul20 Jan21 Jul21 Jan20 Jul20 Jan21 Jul21 Jan20 Jul20 Jan21 Jul21 Jordan Kenya Mexico 100 100 (% change in visitors) Oxford Stringency 50 Google Mobility Index 50 0 −50 0 −100 Jan20 Jul20 Jan21 Jul21 Jan20 Jul20 Jan21 Jul21 Jan20 Jul20 Jan21 Jul21 Somalia Uganda Total 100 100 (% change in visitors) Oxford Stringency Google Mobility 50 Index 50 0 −50 0 −100 Jan20 Jul20 Jan21 Jul21 Jan20 Jul20 Jan21 Jul21 Jan20 Jul20 Jan21 Jul21 Policy Stringency Mobility − transit stations Mobility − workplaces Source: Google, “COVID-19 Community Mobility Reports,” https://www.google.com/covid19/mobility/’ and University of Oxford, “COVID-19 Government Response Tracker,” https://www.bsg.ox.ac.uk/research/covid-19-government-response-tracker. Note: No Google mobility data for Chad, Democratic Republic of Congo, Djibouti, Ethiopia, and Somalia. 71 Figure A2.3 Figure A2.3 Share of Employed by Host and FDP Type, by Country (%) Bangladesh Burkina Faso Chad 100 100 100 50 50 50 0 0 0 Prepandemic Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Prepandemic Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Prepandemic Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Costa Rica Djibouti Ecuador 100 100 100 50 50 50 0 0 0 Prepandemic Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Prepandemic Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Prepandemic Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Ethiopia Iraq Jordan 100 100 100 50 50 50 0 0 0 Prepandemic Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Prepandemic Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Prepandemic Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Kenya Mexico Somalia 100 100 100 50 50 50 0 0 0 Prepandemic Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Prepandemic Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Prepandemic Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Uganda 100 Refugee 50 IDP Hosts 0 Prepandemic Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Source: Staff calculation using HFPS. Note: The figure shows estimates from each survey round, by country. Prepandemic refers to recall questions asking about the period immediately before the pandemic. The periods corresponding to different pandemic stages are shaded in different colors: the prepandemic period is green, April–June 2020 is red, July–December 2020 is blue, and 2021 is gray. January 2020 estimates are based on recall from the earliest available survey wave for each country. Within country estimates use household sample weights. Confidence intervals are shown as vertical lines and are based on heteroskedasticity robust standard errors. 72 Figure A2.4 Share of Households with Respondent Who Stopped Working during the Pandemic, by Host and FDP Type, by Country (%) Bangladesh Burkina Faso Chad 75 75 75 50 50 50 25 25 25 0 0 0 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Costa Rica Djibouti Ecuador 75 75 75 50 50 50 25 25 25 0 0 0 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Ethiopia Jordan Kenya 75 75 75 50 50 50 25 25 25 0 0 0 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Mexico Somalia Uganda 75 75 75 50 50 50 25 25 25 0 0 0 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Refugee IDP Hosts Source: Staff calculation using HFPS. Note: The figure shows estimates from each survey round, by country. The periods corresponding to different pandemic stages are shaded in different colors: the prepandemic period is green, April-June 2020 is red, July-Dec 2020 is blue, and 2021 is gray. January 2020 estimates are based on recall from the earliest available survey wave for each country. Within country estimates use household sample weights. Confidence intervals are shown as vertical lines and are based on heteroskedasticity robust standard errors. 73 A2.5 Share of Households that Received Any Social Assistance Since Pandemic Started, Figure A2.5 Figure by Host and FDP Type, by Country (%) Bangladesh Chad Congo, Dem. Rep. 100 100 100 50 50 50 0 0 0 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Ecuador Ethiopia Iraq 100 100 100 50 50 50 0 0 0 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Jordan Kenya Mexico 100 100 100 50 50 50 0 0 0 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Somalia Uganda 100 100 Refugee IDP 50 50 Hosts 0 0 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Source: Staff calculation using HFPS. Note: The figure shows estimates from each survey round, by country. The periods corresponding to different pandemic stages are shaded in different colors: the prepandemic period is green, April–June 2020 is red, July–December 2020 is blue, and 2021 is gray. Within country estimates use household sample weights. Confidence intervals are shown as vertical lines and are based on heteroskedasticity robust standard errors. 74 Figure A2.6 Share of Households Receiving Assistance during the Pandemic, by Camp Status (%) Burkina Faso Chad Ethiopia 100 100 100 50 50 50 0 0 0 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Iraq Jordan Kenya 100 100 100 50 50 50 0 0 0 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Camped FDPs Uncamped FDPs Hosts Source: Staff estimates using HFPS. Note: Confidence intervals, shown as vertical lines, are based on heteroskedasticity robust standard errors. 75 Figure A2.7 Share of Households that Ran out of Food Because of a Lack of Money or Other Resources Figure A2.730 Days (%) in the Past Burkina Faso Chad 100 100 80 80 60 60 40 40 20 20 0 0 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Congo, Dem. Rep. Costa Rica 100 100 80 80 60 60 40 40 20 20 0 0 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Ecuador Mexico 100 100 80 80 60 60 40 40 20 20 0 0 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Somalia Uganda 100 100 80 80 60 60 40 40 20 20 0 0 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Refugee IDP Hosts Source: Staff calculation using HFPS. Note: The figure shows estimates from each survey round, by country. The periods corresponding to different pandemic stages are shaded in different colors: the prepandemic period is green, April–June 2020 is red, July–December 2020 is blue, and 2021 is gray. Within country estimates use household sample weights. Confidence intervals are shown as vertical lines and are based on heteroskedasticity robust standard errors. 76 Figure A2.8 Share of Households with Household Members Not Eating for a Day due to Lack of Figure A2.8 Resources (%) Burkina Faso Chad Congo, Dem. Rep. 100 100 100 80 80 80 60 60 60 40 40 40 20 20 20 0 0 0 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Costa Rica Djibuti Ecuador 100 100 100 80 80 80 60 60 60 40 40 40 20 20 20 0 0 0 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Kenya Mexico Somalia 100 100 100 80 80 80 60 60 60 40 40 40 20 20 20 0 0 0 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Uganda 100 80 Refugee 60 IDP Hosts 40 20 0 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Source: Staff calculation using HFPS. Note: The figure shows estimates from each survey round, by country. The periods corresponding to different pandemic stages are shaded in different colors: the prepandemic period is green, April–June 2020 is red, July–December 2020 is blue, and 2021 is gray. Within country estimates use household sample weights. Confidence intervals are shown as vertical lines and are based on heteroskedasticity robust standard errors. 77 Figure FigureA2.9 A2.9 Share of Households with Children Accessing Education before and during the Pandemic, by Country and FDP Type Burkina Faso Chad Congo, Dem. Rep. 100 100 100 50 50 50 0 0 0 Prepandemic Prepandemic Prepandemic Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Costa Rica Djibouti Ecuador 100 100 100 50 50 50 0 0 0 Prepandemic Prepandemic Prepandemic Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Ethiopia Iraq Jordan 100 100 100 50 50 50 0 0 0 Prepandemic Prepandemic Prepandemic Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Kenya Mexico Somalia 100 100 100 50 50 50 0 0 0 Prepandemic Prepandemic Prepandemic Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Uganda 100 Refugee IDP 50 Hosts 0 Prepandemic Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Source: Staff calculation using HFPS. Note: The figure shows estimates from each survey round, by country. The periods corresponding to different pandemic stages are shaded in different colors: the prepandemic period is green, April–June 2020 is red, July–December 2020 is blue, and 2021 is gray. Prepandemic refers to recall questions asking about the period immediately before the pandemic. Within country estimates use household sample weights. Confidence intervals are shown as vertical lines and are based on heteroskedasticity robust standard 78 errors. Figure A2.10 Figure A2.10 Share of Households with Children Accessing Education before and during the Pandemic, by Country and Camp Status (%) Burkina Faso Chad Ethiopia 100 100 100 50 50 50 0 0 0 Prepandemic Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Prepandemic Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Prepandemic Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Iraq Jordan Kenya 100 100 100 50 50 50 0 0 0 Prepandemic Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Prepandemic Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Prepandemic Apr20 Jun20 Sep20 Jan21 Jun21 Nov21 Camped FDPs Uncamped FDPs Hosts Source: Staff calculation using HFPS. Note: The figure shows estimates from each survey round, by country. The periods corresponding to different pandemic stages are shaded in different colors: the prepandemic period is green, April–June 2020 is red, July–December 2020 is blue, and 2021 is gray. Coverage is insufficient for Chad wave 2 (March 2021) and is thus omitted from this figure. Prepandemic refers to recall questions asking about the period immediately before the pandemic. Within country estimates use household sample weights. Confidence intervals are shown as vertical lines and are based on heteroskedasticity robust standard errors. 79 Annex 3. Regression Tables Table A3.1a Probability of Working Variable Bangladesh Burkina Faso Chad Costa Rica Djibouti Ecuador -0.209*** -0.145*** 0.018 -0.325*** -0.106 Refugee (0.009) (0.023) (0.04) (0.021) (0.066) -0.276*** IDP (0.015) -0.003 0.003*** 0.004* -0.014 -0.001 -0.003 HH size (0.002) (0.001) (0.002) (0.012) (0.003) (0.008) 0.276*** 0.025 -0.002 0.282*** 0.023 0.353*** Male (0.01) (0.018) (0.028) (0.04) (0.017) (0.033) 0.043*** 0.206*** 0.037 0.032 -0.046* 0 Age above 25 (0.012) (0.06) (0.051) (0.059) (0.027) (0.049) 0.188*** 0.611*** 0.846*** 0.46*** 0.882*** 0.291*** Constant (0.016) (0.061) (0.055) (0.077) (0.033) (0.064) Observations 8162 8497 2327 1592 3864 3088 R-squared .19 .04 .01 .09 .1 .14 Table A3.1a Probability of Working (continued) Variable Ethiopia Iraq Jordan Kenya Mexico Somalia Uganda -0.609*** 0.037 -0.509*** -0.06** -0.503*** Refugee (0.017) (0.033) (0.025) (0.03) (0.023) -0.163*** 0.039 IDP (0.012) (0.038) 0.013 -0.021*** -0.004 0.011*** 0.001 -0.006 0.004 HH size (0.003) (0.002) (0.007) (0.003) (0.007) (0.006) (0.002) 0.175*** 0.478*** 0.069 0.096*** 0.212*** 0.152*** 0.087*** Male (0.018) (0.013) (0.04) (0.014) (0.031) (0.036) (0.014) 0.044 0.171*** -0.113 0.096*** 0.171** 0.08 0.012 Age above 25 (0.027) (0.018) (0.095) (0.018) (0.068) (0.044) (0.034) 0.642*** 0.379*** 0.45 0.586*** 0.418*** 0.329*** 0.815*** Constant (0.032) (0.023) (0.107) (0.021) (0.083) (0.054) (0.035) Observations 8255 8359 1545 41991 2171 4359 7629 R-squared .1 .24 .01 .11 .06 .04 .08 Source: Staff calculation using HFPS. Note: This table reports results based on multivariate OLS regressions, where the dependent variable is a binary indicator for whether the respondent is working or not. The estimated coefficient on the dummy variable indicating whether the respondent is a refugee or IDP measures the difference in the probability of work relative to the national population. Regressions control for household size, gender, and age group (whether respondent is age 25 and above). Within-country samples are weighted using household sample weights. Standard errors are heteroskedasticity robust. 80 Table A3.1b Probability of Income Loss Variable Costa Rica Ecuador Mexico 0.149*** 0.058* 0.164*** Refugee (0.041) (0.03) (0.031) 0.03** 0.011 0.024*** HH size (0.012) (0.008) (0.007) 0.072* 0.013 -0.066** Male (0.042) (0.035) (0.033) 0.35*** 0.575*** 0.335*** Constant (0.055) (0.046) (0.046) Observations 1569 1348 2151 R-squared .02 .02 .1 Source: Staff calculation using HFPS. Note: This table reports results based on multivariate OLS regressions where the dependent variable is a binary indicator for whether the respondent’s household lost income during the pandemic. The coefficient on the dummy variable indicating whether the respondent is a refugee measures the difference in the probability of income loss relative to the national population. Regressions control for household size, gender, and age group (whether respondent is age 25 and above). Within-country samples are weighted using household sample weights. Individual country regressions include survey month fixed effects. Standard errors are heteroskedasticity robust. Table A3.1c Probability of Not Eating for a Day Burkina Costa Variable Chad Djibouti Ecuador Kenya Mexico Somalia Uganda Faso Rica 0.364*** 0.254*** 0.014 0.059 0.155*** 0.35*** 0.402*** Refugee (0.084) (0.027) (0.048) (0.054) (0.014) (0.032) (0.033) 0.122*** 0.105*** IDP (0.008) (0.036) 0 -0.008 0.002 -0.015** 0 0.01** 0.014*** 0.009 0.001 HH size (0.001) (0.014) (0.006) (0.006) (0.006) (0.004) (0.005) (0.006) (0.004) -0.02** -0.099 -0.023 -0.007 -0.033 0.011 0.018 -0.06* 0.039* Male (0.01) (0.098) (0.02) (0.04) (0.028) (0.014) (0.022) (0.035) (0.024) Age above 0.003 0.113 -0.012 0.04 0.08*** -0.01 0.074*** -0.033 -0.007 25 (0.024) (0.071) (0.033) (0.073) (0.018) (0.018) (0.027) (0.044) (0.032) 0.029 0.414*** 0.079** 0.112 0.038 0.016 -0.11** 0.337*** 0.083 Constant (0.025) (0.121) (0.039) (0.085) (0.039) (0.018) (0.044) (0.057) (0.053) Observations 7572 313 1588 473 2921 8444 1722 3640 3663 R-squared .03 .36 .02 .03 .05 .05 .03 .04 .2 Source: Staff calculation using HFPS. Note: This table reports results from multivariate OLS regression where the dependent variable is a binary indicator for whether the household members had not eaten for a day. The coefficient on the dummy variable indicating whether the respondent is a refugee or IDP measures the difference in the probability of not eating for a day relative to the national population. Regressions control for household size, gender, and age group (whether respondent is age 25 and above). Within-country samples are weighted using household sample weights. Individual country regressions include survey month fixed effects. Standard errors are heteroskedasticity robust. 81 Table A3.2 Country-pooled Linear Probability Models on Select Outcomes Currently working Variable Stopped working Ran out of food (refugees) 0.049** 0.048* 0.441*** 0.353*** Refugee (0.02) (0.025) (0.14) (0.105) -0.019 0.05 0.076** 0.245*** IDP (0.02) (0.037) (0.03) (0.028) 0 0 -0.006 0.003 0.007 HH size (0.001) (0.002) (0.007) (0.005) (0.008) 0.012 0.002 0.143*** -0.088** -0.081 Male (0.023) (0.022) (0.035) (0.039) (0.057) 0.013 -0.001 -0.004 -0.037 -0.064 Age above 25 (0.014) (0.016) (0.015) (0.063) (0.075) 0.035* 0.012 Prepandemic work (0.019) (0.017) -0.068* -0.022 Current work (0.042) (0.059) 0.041 -0.018 Agriculture sector (0.027) (0.037) 0.032** -0.081*** -0.575*** Policy Stringency (0.014) (0.013) (0.14) 0.025 -0.021 -0.04 Log of GDP per capita (0.018) (0.068) (0.029) -0.011*** -0.009 -0.084*** GDP per capita growth rate (0.004) (0.008) (0.009) 0.036 Food price inflation (%) (0.024) Restrictiveness of work rights -0.118*** for refugees (0.012) 0.132*** -0.067 0.501 0.385*** -0.023 Constant (0.019) (0.15) (0.54) (0.105) (0.439) Country fixed effects x x Month fixed effects x x Obs 66455 66455 24374 21424 13563 R-squared .06 .04 .2 .22 .25 Bangladesh, Burkina Bangladesh, Chad, Chad, Costa Rica, Faso, Chad, Costa Costa Rica, Djibouti, Ecuador, Mexico, Included countries Rica, Ecuador, Ethiopia, Ecuador, Ethiopia, Uganda Jordan, Kenya, Somalia, Jordan, Kenya, Uganda Mexico, Uganda Source: Staff calculation using HFPS. Note: This table reports results from multivariate OLS regression where the dependent variable is (i) respondent stopped working; (ii) respondent is currently working; and (iii) household ran out of food. The coefficient on the dummy variable indicating whether the respondent is a refugee or IDP measures the difference in the probability of stopping work or running out of food relative to the national population. Regressions control for household size, gender, and age group (whether respondent is age 25 and above). Prepandemic work and current work are dummy variables indicating whether the respondent was working before the pandemic or at the time of survey. The variable agriculture sector is a dummy variable indicating whether the respondent was engaged in the agriculture sector. Policy stringency is measured using data from the Oxford COVID-19 Government Response Tracker (OxCGRT). Data on refugee’s work rights comes from the DWRAP database. Within-country samples are weighted using household sample weights and all countries are equally weighted. Standard errors are heteroskedasticity robust. 82 Annex 4. Estimating Aid for Displaced Populations Using OECD CRS Data Official Development Assistance (ODA) refers to official government financing flows (i.e., aids, loans, and grants) to aid development. The overall aid data come from the OECD’s Creditor Reporting System (CRS) database, which outlines all aid flows made by OECD’s Development Assistance Committee (DAC) member countries,63 non-DAC countries, multilateral organizations, and large private donors.64 Aid flows from OECD donors provide a very close approximation of official ODA figures, representing on average about 95 percent of total ODA (see below). The database is updated annually and is currently available through 2021. Aid to displaced populations is derived using annual CRS disbursement-level data by disaggregating development aid flows intended for displaced populations. Although disbursements are tagged to specific CRS sectors, there are no sector codes dedicated to aiding displacement situations but rather, they are intended to classify various types of humanitarian aid in emergency situations, which includes forced displacement but also natural disaster situations, among others.65 Therefore, the primary strategy here entails keyword extraction based on project identifiers consisting of project titles and descriptions. Any disbursements are counted that include certain keywords, such as refugee, displaced, FDP, returnee, migration, conflict, or UNHCR, and are therefore most likely intended for displaced populations. Estimates counting disbursements tagged to specific CRS sectors are employed as a secondary measure, although the key messages are largely the same. For reference, the indicators tracked under the UNHRC Global Compact on Refugees (GCR) rely on different data sources or methods to proxy refugee financing.66 1.  Coverage of OECD CRS data a. Database includes aid flows by OECD’s DAC member countries, non-DAC countries, multilateral organizations, and large private donors. China is notably excluded. b. ODA numbers from OECD are a reasonably good approximation of the official ODA figures (average 95 percent coverage): 63 OECD, “Development Assistant Committee,” https://www.oecd.org/dac/development-assistance-committee/. 64 OECD, “Creditor Reporting System,” https://www.oecd-ilibrary.org/development/data/creditor-reporting-system_dev-cred-data-en. 65 Relevant sector codes include “Material Relief Assistance and Services (72010),” “Basic Health Care Services in Emergencies (72011),” “Education in Emergencies (72012),” “Emergency Food Assistance (72040),” and “Relief Co-ordination and Support Services (72050).” 66 A dedicated OECD survey on financing refugee situations among members is used to monitor a subset of GCR indicators, such as “Total ODA disbursements from Development Assistance Committee (DAC) donors for the benefit of refugees (and host communities) in developing countries.” For the survey, member countries used their own methods to approximate ODA going to refugee situations. “Total ODA disbursements from DAC donors for the benefit of refugees in developed countries” is estimated with a separate sector code in CRS, and “Number of donors providing official development assistance (ODA) to, or for the benefit of, refugees and host communities in refugee-hosting countries” uses OECD DAC Statistics on Resource Flows to Developing Countries. For details, see UNHCR (2019b) and Hesemann, Desai, and Rockenfeller (2021). 83 Table A4.1 Coverage of OECD CRS Database, Select Countries, 2019 (1) Official ODA 2019, (2) Estimated ODA from OECD Proportion (=(2)/ Country USD millions in 2019 CRS, USD millions in 2019 (1)) Bangladesh 4,483 4,381 98% Burkina Faso 1,149 1,108 96% Chad 707 642 91% Costa Rica 60 56 93% Dem. Rep. of Congo 3,026 2,810 93% Djibouti 272 262 96% Ecuador 525 507 97% Ethiopia 4,810 4,677 97% Iraq 2,212 2,091 95% Jordan 2,797 2,689 96% Kenya 3,251 3,172 98% Mexico 536 525 98% Somalia 1,866 1,720 92% Uganda 2,100 2,028 97% 2.  Estimating aid to displaced populations a.  Using keyword analysis The primary strategy aims to parse out aid flows intended for displaced populations by iteratively extracting keywords from descriptive variables about each aid flow. This includes project titles, short descriptions, and long descriptions. The main keywords are highlighted in the table below: 67 68 Table A4.2 Keywords Used to Estimate Aid for Displaced Populations and Examples Keyword Examples and notes Example 1: “ADVANCING THE RIGHTS AND PROTECTION OF CONFLICT-AFFECTED MIGRANT or OLDER SOUTH SUDANESE MIGRANTS IN ETHIOPIA, UGANDA, AND SOUTH SUDAN” MIGRATION Example 2: “COMMUNITY-LED OUTREACH ON SAFE MIGRATION (COSM)” Example: “SUPPORT TO UNHCR TO PROVIDE INTERVENTIONS IN NUTRITION, HEALTH, UNHCR WATER, AND SANITATION FOR KENYAN REFUGEES AND SUPPORT VOLUNTARY REPATRIATION TO SOMALIA”68 CONFLICT Example: “BUILDING RESILIENCE IN CONFLICT-AFFECTED COMMUNITIES IN IRAQ”69 Example: “ENSURE THE DIGNITY AND QUALITY OF LIFE FOR CONFLICT-AFFECTED INTERNALLY DISPLACED POPULATIONS IN EASTERN DEMOCRATIC REPUBLIC OF THE CONGO” DISPLACED Note: The acronym “FDP” was commonly used to refer to “fertilizer deep placement” technologies. To address this, the use of this acronym was not considered. Instead, the word “displaced” covers nearly all cases used to refer to forcibly displaced persons, including those with the intended use of the acronym. Example: “DOCUMENTATION, SHELTER, AND SOCIAL COHESION FOR INTERNALLY IDP DISPLACED PEOPLE (IDP) AND EDUCATIONAL SUPPORT FOR REFUGEES IN BURKINA FASO” Example: “PSYCHOSOCIAL ASSISTANCE, LIVELIHOODS and DURABLE SOLUTIONS REFUGEE PROGRAM FOR COLOMBIAN REFUGEES IN ECUADOR” RETURNEE Example: “IMPROVING REINTEGRATION OF RETURNEES IN BANGLADESH” 67 All aid flows that were channeled through UNHCR regardless of project descriptions were also included. 68 The term “conflict” is commonly used to refer to governance projects involving institutional conflict and therefore the term is 84 used only when used in conjunction with the terms “violent,” “victim,” or “affected.” b.  Using sector codes Each ODA transaction is tagged to a standardized set of sector codes specified in the CRS. To assess the robustness of the primary strategy using keyword extraction, a secondary indicator of aid for displaced populations was also considered: the emergency response sector. This exploits subcategories of the three- digit sector code 720 in the CRS codebook whose description most closely matches projects intended for displaced populations and are listed in Table A4.3.69 Emergency response mainly includes assistance in times of crises and subsequent rehabilitation. Compared to the primary approach of key word analysis, this secondary measure is qualitatively less precise in identifying aid targeting displaced groups. Although there is reasonable correlation for most country groups between the estimates based on the two different methods, the secondary method tends to produce more volatile estimates over time. Similar to estimates derived using the keyword search method, total disbursements intended for displaced populations fell from their 2019 peak, globally as well as in the IDA-18 RSW and GCFF countries and in major hosting countries (Figure A3.1). There is a strong correlation in the estimated annual aid totals for the sample of HFPS countries and the IDA-18 RSW and GCFF countries through 2020. Table A4.3 Detailed Sector Codes in the OECD CRS Emergency Response Sector Shelter, water, sanitation, education, health services, including supply of medicines, and malnutrition management, including medical nutrition management; supply of other nonfood relief items (including cash and voucher delivery modalities) for the Material Relief benefit of crisis-affected people, including refugees and IDPs in developing countries. Assistance Includes assistance delivered or coordinated by international civil protection units and Services in the immediate aftermath of a disaster (in-kind assistance, deployment of specially (72010) equipped teams, logistics and transportation, or assessment and coordination by experts sent to the field). Also includes measures to promote and protect the safety, well-being, dignity, and integrity of crisis-affected people, including refugees and IDPs in developing countries. Basic Health Provision of health services (basic health services, mental health, sexual and Care Services reproductive health), medical nutritional intervention (therapeutic feeding and medical in Emergencies interventions for treating malnutrition), and supply of medicines for the benefit of (72011) affected people. Excludes supplemental feeding (72040) Support for education facilities (including restoring preexisting essential infrastructure and school facilities), teaching, training and learning materials (including digital technologies, as appropriate), and immediate access to quality basic and primary Education in education (including formal and non-formal education), and secondary education Emergencies (including vocational training and secondary level technical education) in emergencies (72012) for the benefit of affected children and youth, particularly targeting girls and women and refugees, life skills for youth and adults, and vocational training for youth and adults Provision and distribution of food; cash and vouchers for the purchase of food; non- Emergency medical nutritional interventions for the benefit of crisis-affected people, including Food refugees and IDPs in developing countries in emergency situations. Includes logistical Assistance costs. Excludes non-emergency food assistance (52010), food security policy and (72040) administrative management (43071), household food programs (43072), and medical nutrition interventions (therapeutic feeding) (72010 and 72011). 69  In 2020, material relief assistance and services accounted for 56 percent of emergency response disbursements, while emergency food assistance and relief coordination accounted for 25.3 percent and 18.6 percent, respectively. During the sample period 2016–20, there was no disbursement made for purposes 72011 (Basic Health Care Services in Emergencies) and 72012 (Education in Emergencies). 85 Measures to coordinate the assessment and safe delivery of humanitarian aid, Relief Co- including logistic, transport, and communication systems; direct financial or technical ordination support to national governments of affected countries to manage a disaster situation; and Support activities to build an evidence base for humanitarian financing and operations, sharing Services this information and developing standards and guidelines for more effective response; (72050) and funding for identifying and sharing innovative and scalable solutions to deliver effective humanitarian assistance Source: OECD, “DAC and CRS Code Lists,” https://www.oecd.org/dac/financing-sustainable-development/development-finance- standards/dacandcrscodelists.htm. Table A4.4 Top ten recipient countries in terms of aid for displaced populations, 2016-2021 (in 2020 million $US) Year Country Aid for displaced populations 2016 Syrian Arab Republic 1,130 2016 Türkiye 765 2016 Jordan 683 2016 Iraq 616 2016 Lebanon 568 2016 West Bank and Gaza 543 2016 Burundi 349 2016 South Sudan 185 2016 Nigeria 172 2016 Yemen, Rep. 148 2017 Syrian Arab Republic 982 2017 Türkiye 972 2017 Iraq 577 2017 Lebanon 560 2017 Jordan 529 2017 South Sudan 443 2017 Uganda 236 2017 Libya 213 2017 Bangladesh 198 2017 Somalia 192 2018 Syrian Arab Republic 802 2018 Iraq 773 2018 Türkiye 762 2018 Jordan 682 2018 Lebanon 596 2018 Bangladesh 410 2018 Ethiopia 289 2018 Yemen, Rep. 222 2018 Uganda 191 2018 Somalia 190 2019 Türkiye 1,060 2019 Syrian Arab Republic 692 2019 Jordan 607 2019 Lebanon 566 86 2019 Iraq 525 2019 Bangladesh 505 2019 Ethiopia 253 2019 Yemen, Rep. 251 2019 Uganda 243 2019 South Sudan 240 2020 Türkiye 858 2020 Syrian Arab Republic 648 2020 Lebanon 509 2020 Jordan 499 2020 Bangladesh 449 2020 Iraq 424 2020 South Sudan 236 2020 Ethiopia 229 2020 Uganda 224 2020 Colombia 207 2021 Türkiye 1,285 2021 Colombia 1,224 2021 Jordan 570 2021 Lebanon 460 2021 Bangladesh 425 2021 Iraq 404 2021 Syrian Arab Republic 354 2021 Ethiopia 309 2021 West Bank and Gaza 278 2021 South Sudan 271 Source: Staff calculations using OECD CRS disbursement data. Figure A4.1 Aid A4.1 to Displaced Populations, Globally and by Country Groupings (in 2020 million US$) 12,000 US$ million (2020 constant) Global 7,000 IBRD/IDA major hosts IDA−18 RSW & GCFF Countries in HFPS sample 2,000 2016 2017 2018 2019 2020 2021 Source: Staff calculations using “DAC and CRS Code Lists,” https://www.oecd.org/dac/financing-sustainable-development/ development-finance-standards/dacandcrscodelists.htm. Note: Global aid flows correspond to the secondary axis on the right. 87 Figure A4.2. Total Aid by country (in 2020 million US$) Figure A4.2 Bangladesh Burkina Faso Chad 10,000 1,500 1,000 150 2,000 100 100 1,000 1,500 50 50 1,000 500 0 0 0 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021 Congo, Dem. Rep. Costa Rica Djibouti 4,000 2,000 400 10 10 200 300 200 2,000 0 0 0 0 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021 Ecuador Ethiopia Iraq 6,000 4,000 1,000 2,000 3,000 800 300 30 600 200 20 4,000 2,000 400 100 10 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021 Jordan Kenya Mexico 5,000 6,000 3,000 4,000 5,000 700 40 140 4,000 4,000 600 120 20 100 3,000 2,000 500 0 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021 Somalia Uganda 4,000 200 2,000 3,000 200 3,000 Total aid to displaced populations Total aid disbursed (R) 2,000 1,000 100 0 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021 Source: Staff calculations using “DAC and CRS Code Lists,” https://www.oecd.org/dac/financing-sustainable-development/ development-finance-standards/dacandcrscodelists.htm. Note: Total disbursements correspond to the secondary axis on the right. 88 Figure A4.3 Figure A4.3. Aid per displaced person by country (in 2020 US$) Bangladesh Burkina Faso Chad 600 150 100 150 200 400 100 200 50 50 0 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021 Congo, Dem. Rep. Costa Rica Djibouti 400 400 40 200 300 200 30 100 20 0 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021 Ecuador Ethiopia Iraq 400 200 150 100 200 100 50 0 0 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021 Jordan Kenya Mexico 400 600 800 1000 300 300 200 200 100 100 0 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021 Somalia Uganda 200 200 150 Aid to displaced populations (per displaced person) 100 Total aid (per capita) 50 0 2016 2017 2018 2019 2020 2021 2016 2017 2018 2019 2020 2021 Source: Staff calculations using “DAC and CRS Code Lists,” https://www.oecd.org/dac/financing-sustainable-development/ development-finance-standards/dacandcrscodelists.htm. 89 References Adebisi Yusuff Adebisi, Kirinya Ibrahim, Don Forced Displacement, Copenhagen. https:// Eliseo Lucero-Prisno, Aniekan Ekpenyong, www. jointdatacenter.org/covid-19-impact- Alumuku Iordepuun Michael, Iwendi Godsgift monitoring-on-refugee-households-in-chad/; Chinemelum, and Ayomide Busaya Sina- https://www. jointdatacenter.org/covid-19- Odunsi. 2019. “Prevalence and Socio-economic impact-monitoring-on-refugee-households-in- Impacts of Malnutrition Among Children in chad-brief-no-2/. Uganda.” Nutrition and Metabolic Insights 12. Barchfield, Jenny. 2021. “Pandemic Deepens 10.1177/1178638819887398. Hunger for Displaced People the World Over.” Ahmad, Yasmin, and Eleanor Carey. 2021. “How Stories, United Nations High Commissioner COVID-19 and Russia’s War of Aggression against for Refugees, March 31. https://www.unhcr. Ukraine are Reshaping Official Development org/news/stories/pandemic-deepens-hunger- Assistance (ODA).” Development Co-operation displaced-people-world-over. Profiles, Organisation for Economic Co- Beegle, Kathleen, and Luc Christiaensen, eds. operation and Development, Paris. 2019. Accelerating Poverty Reduction in Africa. Alan, Sule, Ceren Baysan, Mert Gumren, and Washington, DC: World Bank. Elilf Kubilay. 2021. “Building Social Cohesion Blair, Christopher W., Guy Grossman, and Jeremy in Ethnically Mixed Schools: An Intervention M. Weinstein. 2021. “Forced Displacement on Perspective Taking.” Quarterly Journal of and Asylum Policy in the Developing World.” Economics 136 (4): 2147–94. International Organization 76 (2): 337–78. Ambel, Alemayehu, Kevin McGee, and Asmelash Brubaker, Joshua, Talip Kilic, and Philip Wollburg. Tsegay. 2021. “Reducing Bias in Phone Survey 2021. “Representativeness of Individual-Level Samples: Effectiveness of Reweighting Data in COVID-19 Phone Surveys: Findings from Techniques Using Face-to-Face Surveys as Sub-Saharan Africa.” PloS ONE 16 (11). Frames in Four African Countries.” Policy Brunckhorst, Ben, Alexandru Cojacaru, Ruth Hill, Research Working Paper 9676, World Bank, Yeon Soo Kim, and Maurice Kugler. 2023. Washington, DC. “Long COVID: The Evolution of Household Arena, Marco, Emilio Fernandez Corugedo, Welfare in Developing Countries During the Jaime Guajardo, and Juan Francisco Yepez. Pandemic.” Policy Research Working Paper 2022. “Venezuela’s Migrants Bring Economic 10300, World Bank, Washington, DC. https:// Opportunity to Latin America.” IMF Country openknowledge.worldbank.org/server/api/ Focus (blog), December 7. https://www.imf.org/ core/bitstreams/337532ce-e319-49e9-af9a- en/News/Articles/2022/12/06/cf-venezuelas- 3ffe4fe59039/content. migrants-bring-economic-opportunity-to-latin- Brunckhorst, Ben, Alexandru Cojocaru, and Yeon america. Soo Kim. Forthcoming. “Tracing Pandemic Ballivian, Amparo, João Pedro Azevedo, Will Durbin, Impacts in the Absence of Regular Survey Data: Jesus Rios, Johanna Godoy, Christina Borisova, What Have We Learned from the World Bank’s and Sabine Mireille Ntsama. 2013. “Listening to High-Frequency Phone Surveys?” World Bank, LAC: Using Mobile Phones for High Frequency Washington, DC. Data Collection. Final Report.” World Bank, Bundervoet, Tom, Maria E. Davalos, and Natalia Garcia. Washington, DC. 2021. “The Short-Term Impacts of COVID-19 Baradine, Zakaria, Ahmet Abderahim Abbo, Cherif on Households in Developing Countries. An O. A. Haggar, A. Ibrahim Abdallah, Aboudrahyme Overview Based on a Harmonized Data Set Savadogo, Sidi M. Sawadogo, Harriet Mugera, of High-Frequency Surveys.” Policy Research Jeffery Tanner, and Sosso Feindouno. 2021. Working Paper 9582, World Bank, Washington, “COVID-19 Impact Monitoring on Refugee DC. https://openknowledge.worldbank.org/ Households in Chad.” Results from a High- server/api/core/bitstreams/9e310f88-7152-5f44- Frequency Phone Survey–Refugee Round 1: 82be-a0ef7c280c5d/content. Jan/Feb 2021, Briefs 1–2. Joint Data Center on 90 CCCM (Camp Coordination and Camp Management) MA. https://www.hsph.harvard.edu/wp- Cluster Somalia. 2021. “CCCM Cluster Somalia content/uploads/sites/2464/2018/05/UNHCR- Dashboard as of November 2021.” Operational ECUADOR-Report1.pdf. Data Portal, United Nations High Commissioner D’Souza, Anna, Eliana Favari, Siddharth for Refugees, Geneva. https://data2.unhcr.org/ Krishnaswamy, and Sharad Tandon. 2022. en/documents/details/90057. “Consequences of Forced Displacement in CCCM, REACH, UNHCR (United Nations High Active Conflict: Evidence from the Republic Commissioner for Refugees). 2021. “Iraq: IDP of Yemen.” Policy Research Working Paper Camp Profiling, Camp Directory – Round XV, 10176, World Bank, Washington, DC. https:// June-August 2021.” ReliefWeb, March 2. https:// openknowledge.worldbank.org/server/api/ reliefweb.int/report/iraq/iraq-idp-camp-profiling- core/bitstreams/4284de4d-4145-5551-8313- camp-directory-round-xv-june-august-2021. af6b1418c042/content. CONASUR (Conseil National de Secours d’Urgence EGRISS (Expert Group on Refugee, Internally et de Réhabilitation) and UNHCR. 2021. Displaced Persons and Statelessness Statistics). “Enregistrement des Personnes Deplacees 2023a. “Compilers’ Manual on Forced Internes du Burkina Faso.” Operational Displacement Statistics.” EGRISS, Copenhagen. Data Portal, UNHCR, Geneva. https://data. https://egrisstats.org/wp-content/uploads/2023- globalprotectioncluster.org/en/documents/ EGRISS_CM_Final-web.pdf. details/578. ———. 2023b. “Towards a Standardized Approach to Corral, Paul, Alexander Irwin, Nandini Krishnan, Identify IDPs, Refugees and Related Populations Daniel Gerszon Mahler, and Tara Vishwanath. in Household Surveys.” Methodological 2020. Fragility and Conflict: On the Front Lines Paper Series 1, EGRISS, Copenhagen. https:// of the Fight against Poverty. Washington, DC: egrisstats.org/wp-content/uploads/EGRISS- World Bank. Methodological-Paper-Towards-a-standardized- Dabalen, Andrew, Alvin Etang, Johannes approach-to-identify-IDPs-refugees-1.pdf. Hoogeveen, Elvis Mushi, Youdi Schipper, Egger, Dennis, Edward Miguel, Shana S. Warren, and Johannes von Engelhardt. 2016. Mobile Ashish Shenoy, Elliott Collins, Dean Karlan, Doug Phone Panel Surveys in Developing Countries: Parkerson et al. 2021. “Falling Living Standards A Practical Guide for Microdata Collection. during the COVID-19 Crisis: Quantitative Washington, DC: World Bank. Evidence from Nine Developing Countries.” Dang, Hai-Anh H., and Paolo Verme. 2022. Science Advances 7 (6). “Estimating Poverty for Refugees in Data-Scarce Favari, Eliana, Michael Geiger, Siddharth Krishnas- Contexts: An Application of Cross-Survey wamy, and Sharad Tandon. 2021. “The 2020 Imputation.” Journal of Population Economics Food Security Crisis in Yemen.”  World Bank 36: 653–79. and World Food Programme, Washington, Davis, Austin C., Mohammed Masudur Rahman, DC and Geneva. https://thedocs.worldbank. Nodoka Hasegawa, Paola Cadoni, and Theresa org/en/doc/70c4a9845ac4bc5f5a857cbb- Beltramo. 2022. “The Impact of Financial 943054de-0280012022/original/Food-Securi- Assistance through Volunteer Programmes in ty-Crisis.pdf. Cox’s Bazar Refugee Camps.” United Nations FSIN (Food Security Information Network), and High Commissioner for Refugees, Geneva. Global Network against Food Crises. 2022. Brief on impact of financial assistance through “2022 Global Report on Food Crises: Joint volunteer programmes.pdf. Analysis for Better Decisions.” World Food Dempster, Helen, Thomas Ginn, Jimmy Graham, Programme, Rome. https://www.wfp.org/ Martha Guerrero Ble, Daphne Jayasinghe, and publications/global-report-food-crises-2022. Barri Shorey. 2020. “Locked Down and Left Garlick, Robert, Kate Orkin, and Simon Quinn. Behind: The Impact of COVID-19 on Refugees’ 2015. “Call Me Maybe: Experimental Evidence Economic Inclusion.” Policy Paper 178, Center for on Using Mobile Phones to Survey African Global Development, Washington, DC. https:// Microenterprises.” CSAE Working Paper www.cgdev.org/sites/default/files/locked-down- WPS/2016-14, Center for the Study of African and-left-behind-paper-71320.pdf. Economies, University of Oxford. Devarajan, Shantayanan. 2013. “Africa’s Statistical GCFF (Global Concessional Financing Facility). Tragedy.” Review of Income and Wealth 59 (S1): 2021. “Global Concessional Financing Facility S9–S15. 2020–2021 Annual Report.” GCFF, World Donger, Elizabeth, Arlan Fuller, Jacqueline Bhabha, Bank, Washington, DC. https://globalcff.org/wp- and Jennifer Leaning. 2017. “Protecting Refugee content/uploads/2021/10/FINAL_WBG-GCFF- Youth in Ecuador: An Evaluation of Health 2021-Annual-ReportPage.pdf. and Wellbeing.” FXB Center for Health and Ginn, Thomas. 2023. “Labor Market Access and Human Rights, Harvard University, Cambridge, Outcomes for Refugees.” JDC Quarterly Digest 91 on Forced Displacement, Seventh Issue. UNHCR Kenya, Government of. 2022. “2021/2022 Joint Data Center and World Bank, Geneva Programme Based Budget of the National and Washington, DC. https://doi.org/10.47053/ Government of Kenya for the Year Ending jdc.300123. 30th June 2022.” Republic of Kenya, Nairobi. Ginn, Thomas, Reva Resstack, Helen Dempster, h t t p s : / / w w w.t r e a s u r y. g o . ke / w p - c o n t e n t / Emily Arnold-Fernández, Sarah Miller, Martha uploads/2021/07/PBB_July2021_Approved.pdf. Guerrero Ble, and Bahati Kanyamanza. 2022. Kirolos,  Amir,  Magdalena Goyheneix,  Mike Kalmus “2022 Global Refugee Work Rights Report.” Eliasz, Mphatso Chisala, Samantha Lissauer, Center for Global Development, Washington, Melissa Gladstone, and Marko Kerac. 2022. DC. https://www.cgdev.org/publication/2022- “Neurodevelopmental, Cognitive, Behavioural global-refugee-work-rights-report. and Mental Health Impairments Following Grantham-McGregor, Sally. 1995. “A Review of Childhood Malnutrition: A Systematic Review.” Studies of the Effect of Severe Malnutrition BMJ Global Health 7 (7). on Mental Development.”  The Journal of Lahn, Glada, and Owen Grafham. 2015. “Heat, Light Nutrition 125 (suppl. 8): 2233S–38S. and Power for Refugees Saving Lives, Reducing Grosh, Margaret E., and Paul Glewwe. 1998. “Data Costs.” Chatham House Report, Royal Institute Watch: The World Bank’s Living Standards of International Affairs, London. Measurement Study Household Surveys.” Mahler, Daniel Gerszon, Nishant Yonzan, and Journal of Economic Perspectives 12 (1): 187–96. Christoph Lakner. 2022. “The Impact of Hesemann, Jens, Harsh Desai, and Yasmine COVID-19 on Global Inequality and Poverty.” Rockenfeller. 2021. “Financing for Refugee Paper prepared for the IARIW-TNBS Conference, Situations 2018-19.” Forced Displacement “Measuring Income, Wealth and Well-Being in Series. Organisation for Economic Co-operation Africam,” Arusha, Tanzania, November 11–13. and Development, Paris. https://www.oecd.org/ https://iariw.org/wp-content/uploads/2022/10/ dac/conflict-fragility-resilience/docs/financing- Mahler-et-al-IARIW-TNBS-2022.pdf. refugee-situations-2018-19.pdf. Malaeb, Bilal, Anne Duplantier, Romeo Jacky Gansey, Himelein, Kristen. 2014. “Weight Calculations for Sekou Tidani Konate, Omar Abdoulkader, Jeff Panel Surveys with Subsampling and Split-off Tanner, and Harriet Mugera. 2021. “Monitoring Tracking.” Statistics and Public Policy 1 (1): 40-45. the Socio-Economic Impacts of COVID-19 on DOI: 10.1080/2330443X.2013.856170. Djiboutian and Refugee Households in Djibouti: Himelein, Kristen, Stephanie Eckman, Kevin McGee, Results from the Third Wave of Survey.” World Michael Wild, and Nobuo Yoshida. 2020. “High Bank, Washington, DC. http://hdl.handle. Frequency Mobile Phone Surveys of Households net/10986/35702. to Assess the Impacts of COVID-19: Guidelines Martins, Vinicius J., Telma M. M. Toledo Florêncio, on Sampling Design.” World Bank, Washington, Luciane P. Grillo, Maria do Carmo P. Franco, DC. Paula A. Martins, Ana Paula G. Clemente, Carla IDMC (Internal Displacement Monitoring Centre). D. L. Santos, Maria de Fatima A. Vieira, and 2020. “GRID 2020: Global Report on Internal Ana Lydia Sawaya. 2011. “Long-Lasting Effects Displacement.” IDMC, Norwegian Refugee of Undernutrition.” International Journal of Council, Geneva. https://www.internal- Environmental Research and Public Health 8 displacement.org/global-report/grid2020/. (6): 1817–46. ———. 2021. “GRID 2021. Global Report on Internal OCHA (Office for the Coordination of Humanitarian Displacement 2021.” IDMC, Norwegian Refugee Affairs). 2021. “République Démocratique Council, Geneva. du Congo: Personnes Déplacées Internes ———. 2022. “GRID 2022: Global Report on et Retournées.” OCHA, New York. https:// Internal Displacement.” IDMC, Norwegian reliefweb.int/report/democratic-republic-congo/ Refugee Council, Geneva. https://www.internal- r-publique-d-mocratique-du-congo-personnes- displacement.org/global-report/grid2022/. d-plac-es-internes-et-22. IMF (International Monetary Fund). 2020. “Policies OECD (Organisation for Economic Co-operation to Support People during the Pandemic.” Fiscal and Development). 2021. International Migration Monitor (April), IMF, Washington, DC. Outlook 2021. OECD, Paris. IOM (International Organization for Migration). ———. 2022. International Migration Outlook 2022. 2020. “Iraq: Access to Durable Solutions Among OECD, Paris. IDPs in Iraq. Five Years in Displacement.” IOM, Olivieri, Sergio, Francesc Ortega, Ana Rivadeneira, Geneva. and Eliana Carranza. 2020. “Shoring Up ITU (International Telecommunication Union). 2021. Economic Refugees. Venezuelan Migrants in Measuring Digital Development: Facts and the Ecuadoran Labor Market.” Policy Research Figures 2021. Geneva: ITU. https://www.itu.int/ Working Paper 9332, World Bank, Washington, itu-d/reports/statistics/facts-figures-2021/. DC. https://documents1.worldbank.org/curated/ 92 en/354821595510754331/pdf/Shoring-Up- Schady, Norbert, Alaka Holla, Shwetlana Sabarwal, Economic-Refugees-Venezuelan-Migrants-in- Joana Silva, and Andres Yi Chang. 2023. the-Ecuadoran-Labor-Market.pdf. Collapse and Recovery. How the COVID-19 Pape, Utz J., and Luca Parisotto. 2019. “Estimating Pandemic Eroded Human Capital and What to Poverty in a Fragile Context: The High Frequency Do about It. Washington, DC: World Bank. Survey in South Sudan.” Policy Research Working Schuettler, Kirsten, and Laura Caron. 2020. “Jobs Paper 8722, World Bank, Washington, DC. Interventions for Refugees and Internally https://openknowledge.worldbank.org/server/ Displaced Persons.” Jobs Working Paper 47, api/core/bitstreams/b3495185-4703-5f81-a170- World Bank, Washington, DC. http://hdl.handle. de443894f002/content. net/10986/33953. Pape, Utz J., and Philip Wollburg. 2019. “Estimation Sharma, Dhiraj, and Matthew Wai-Poi. 2019. of Poverty in Somalia Using Innovative “Arrested Development: Conflict, Displacement, Methodologies.” Policy Research Working Paper and Welfare in Iraq.” World Bank, Washington, 8735, World Bank, Washington, DC. https:// DC. https://documents1.worldbank.org/curated/ openknowledge.worldbank.org/server/api/ en/914571554209012533/pdf/Arrested- core/bitstreams/35514d7f-6d1a-5891-9933- Development-Conflict-Displacement-and- ce67d06c11d4/content. Welfare-in-Iraq.pdf. Pape, Utz J., Antonia Delious, Ritika Khandelwal, Stave, Svein Erik, and Solveig Hillesund. 2015. and Rhea Gupta. 2020. “Socioeconomic Impact Impact of Syrian Refugees on the Jordanian of COVID-19 in Kenya. November.” World Bank, Labour Market. Findings from the Governorates Washington, DC. https://documents.worldbank. of Amman, Irbid and Mafraq. Geneva: org/en/publication/documents-reports/ International Labour Organization. documentdetail/384651613652984513/socio- Tanner, Jeffery C. 2021. “The Pitfalls and Potential of economic-impacts-of-covid-19-in-kenya-results- High-Frequency Phone Surveys during COVID-19.” update. Forced Migration Review 66: 55–57. https:// ———. 2021. “Socioeconomic Impacts of COVID-19 ieg.worldbankgroup.org/sites/default/files/ in Kenya. June.” World Bank, Washington, Data/Evaluation/files/ImpactsofInterventions_ DC. https://openknowledge.worldbank.org/ EarlyChildhoodonLaterOutcomes.pdf entities/publication/d9b39f93-7462-5421-81de- Tanner, Jeffery C., Tara Candland, and Whitney S. 3165dee064bd. Odden. 2015. Later Impacts of Early Childhood Pape, Utz J., Antonia Delious, Ritika Khandelwal, Interventions: A Systematic Review. IEG Working Rhea Gupta, and Alastair Haynes. 2021. Paper 2015/3, Independent Evaluation Group. “Socioeconomic Impact of COVID-19 in World Bank Group, Washington, DC. https:// Kenya. April.” World Bank, Washington, ieg.worldbankgroup.org/sites/default/files/ DC. https://documents.worldbank.org/ Data/Evaluation/files/ImpactsofInterventions_ en/publication/documents-reports/ EarlyChildhoodonLaterOutcomes.pdf documentdetail/434961617772887975/ Tanner, Jeffery, Harriet Mugera, Domenico socioeconomic-impacts-of-covid-19-in-kenya. Tabasso, Maja Lazić, and Björn Gillsäter. Pape, Utz J., Antonia Delious, Mirko Vintar, Rhea 2021. “Answering the Call: Forcibly Displaced Gupta, Alastair Haynes, Abril Rios Riviera, Eugenie during the Pandemic.” JDC Paper Series on Rose Fontep, Theresa Beltramo, and Florence Forced Displacement 2, Joint Data Center on Nimoh. 2021. “How COVID-19 Continues to Affect Forced Replacement, United Nations High Lives of the Refugees in Kenya. Rapid Response Commissioner for Refugees, Geneva. https:// Phone Surveys Rounds 1 to 5.” World Bank, www.unhcr.org/answering-the-call.html. Washington, DC. https://documents.worldbank. Tiberti, Marco, Clarence Tsimpo Nkenge, Marco org/en/publication/documents-reports/ Costantini, Harriet Mugera, Jeff Tanner, documentdetail/202201637042522937/how- and Zakaria Koncobo. 2021. “Burkina Faso. covid-19-continues-to-affect-lives-of-refugees- COVID-19 Impact Monitoring of Internally in-kenya-rapid-response-phone-survey-rounds- Displaced Persons.” Bulletin Nos. 1–3, May– 1-to-5. July. Joint Data Center, UNHCR, Geneva. Rodriguez, Laura, and Emma Smith. Forthcoming. https://www.jointdatacenter.org/wp-content/ “The Impact of the Covid-19 Pandemic on uploads/2021/12/Brief-Round1-IDPs_final_ Host Communities and Refugees in Jordan.” En.pdf; https://www.jointdatacenter.org/wp- Joint Data Center on Forced Displacement, content/uploads/2022/04/BFA_COVID19_ Copenhagen. Round10-IDPs-English_final_21Mar2022.pdf; Rudin-Rush, Lorin, Jeffrey D. Michler, Anna https://www.jointdatacenter.org/wp-content/ Josephson, and Jeffrey R. Bloem. 2022. “Food uploads/2022/08/Burkina-Faso_brief-no3.pdf. Insecurity during the First Year of the COVID-19 Uganda, Government of. 2021. “Uganda Refugees Pandemic in Four African Countries.”  Food and Asylum Seekers as of 30 November 2021.” Policy 111: 102306. Operational Data Portal, United Nations High 93 Commissioner for Refugees, Geneva. https:// ———. 2021g. “UNHCR’s Most Underfunded data.unhcr.org/en/documents/details/90196. Situations in 2021.” UNHCR, Geneva. UN (United Nations). 2020. “COVID-19 Worsening ———. 2022a. “All Inclusive: The Campaign for Food Insecurity, Driving Displacement, Warn Refugee Education.” UNHCR, Geneva. https:// UN Agencies.” UN News, November 10. https:// www.unhcr.org/media/unhcr-education-report- news.un.org/en/story/2020/11/1077272. 2022-all-inclusive-campaign-refugee-education. UNDP (United Nations Development Programme). ———. 2022b. “Annual Results Report 2022: 2017. “Uganda’s Contribution to Refugee Jordan.” UNHCR, Geneva. https://reporting. Protection and Management.” Summary of unhcr.org/operational/operations/jordan. Study, UNDP, New York. https://www.undp.org/ ———. 2022c. “Assessing the Socioeconomic Impact uganda/publications/ugandas-contribution- of COVID-19 on Forcibly Displaced Populations. refugee-protection-and-management. Thematic Brief No. 4: The Case of Costa Rica.” UNESCO. 2019. Migration, Displacement and UNHCR, Geneva. https://www.unhcr.org/sites/ Education: Building Bridges Not Walls. Global default/files/legacy-pdf/624bf06c4.pdf. Education Monitoring Report. Paris: UNESCO. ———. 2022d. “Ecuador.” Monthly Update, July– UNHCR (United Nations High Commissioner August. UNHCR, Geneva. https://reporting. for Refugees). 2016. “Left Behind: Refugee unhcr.org/document/3260. Education in Crisis.” UNHCR, Geneva. ———. 2022e. “Education Access for the Forcibly ———. 2019a. “Africa.” In “UNHCR Global Report Displaced During the Pandemic.” UNHCR, 2019.” UNHCR, Geneva. https://reporting.unhcr. Geneva. https://www.unhcr.org/sites/default/ org/sites/default/files/gr2019/pdf/03_Africa.pdf. files/2023-01/63cabf374.pdf. ———. 2019b. “Global Compact on Refugees: ———. 2022f. “Global Trends: Forced Displacement Indicator Framework.” UNHCR, Geneva. https:// in 2021.” UNHCR, Geneva. https://www.unhcr. reliefweb.int/report/world/global-compact- org/62a9d1494/global-trends-report-2021. refugees-indicator-framework-july-2019. UNHCR, and Bangladesh, Government of. 2021. ———. 2019c. “Stepping Up: Refugee Education “Joint Government of Bangladesh - UNHCR in Crisis.” UNHCR, Geneva. https://www. Population Factsheet as of 30 November 2021.” unhcr.org/steppingup/wp-content/uploads/ Rohingya Refugee Response/Bangladesh, sites/76/2019/09/Education-Report-2019-Final- Operational Data Portal, UNHCR, Geneva. https:// web-9.pdf. data.unhcr.org/en/documents/details/90045. ———. 2020. “Coming Together for Refugee UNHCR, and CNR (Commission Nationale pour Education.” UNHCR, Geneva. https://www.unhcr. les Refugies). 2021. “Democratic Republic of org/media/coming-together-refugee-education- the Congo: Monthly Statistics of Refugees and education-report-2020. Asylum Seekers as of 30 November 2021.” ———. 2021a. “Ethiopia: Refugees and Asylum- Operational Data Portal, UNHCR, Geneva. https:// Seekers as of 30 November 2021.” Operational data.unhcr.org/en/documents/details/90048. Data Portal, UNHCR, Geneva. https://data.unhcr. UNHCR, and WFP (World Food Programme). 2020. org/en/documents/details/90203. “UNHCR and WFP Warn Refugees in Africa Face ———. 2021b. “Global Trends: Forced Displacement Hunger and Malnutrition as COVID-19 Worsens in 2020.” UNHCR, Geneva. https://www.unhcr. Food Shortages.” Press Release, July 9. https:// org/statistics/unhcrstats/60b638e37/global- www.unhcr.org/us/news/news-releases/unhcr- trends-forced-displacement-2020.html. and-wfp-warn-refugees-africa-face-hunger-and- ———. 2021c. “Iraq: Syrian Refugee Statistics, malnutrition-covid-19-worsens. November 2021.” Operational Data Portal, UNHCR, UNICEF, and WFP (World Food Programme). UNHCR, Geneva. https://data.unhcr.org/en/ 2020. “Multi-sectoral Rapid Needs Assessment: documents/details/90154. COVID19 - Jordan.” UNHCR, Geneva. https:// ———. 2021d. “Kenya: Registered Refugees and data.unhcr.org/es/documents/details/76215. Asylum Seekers as of 31 December 2021.” UNICEF (United Nations Children’s Fund). 2022. Operational Data Portal, UNHCR, Geneva. https:// “Education, Children on the Move and Inclusion data.unhcr.org/en/documents/details/90423. in Education.” UNICEF, New York. https://www. ———. 2021e. “Registered Persons of Concern unicef.org/reports/education-children-move- Refugees and Asylum Seekers in Jordan.” and-inclusion-education. Operational Data Portal, UNHCR, Geneva. https:// Vishwanath, Tara, Arthur Alik-Lagrange, and data.unhcr.org/en/documents/details/90152. Leila Aghabarari. 2020. “Highly Vulnerable ———. 2021f. “UNHCR TCHAD: “Personnes Yet Largely Invisible. Forcibly Displaced in Relevant de la Compétence du HCR Novembre the COVID-19-Induced Recession.” Joint Data 2021.” Operational Data Portal, UNHCR, Center on Forced Displacement, United Nations Geneva. https://data.unhcr.org/en/documents/ High Commissioner for Refugees, Geneva. details/90043. https://www.jointdatacenter.org/wp-content/ 94 uploads/2020/12/JDC-Paper-Series-on-Forced- Refugee Households in Chad.” Results from a Displacement_No.1_Final.pdf. High-Frequency Phone Survey – Refugee Round Wagner, Emma, and Sebastien Hine. 2021. “Progress 1: Jan/Feb 2021, World Bank, Washington, DC. under Threat. Refugee Education One Year on https://www.jointdatacenter.org/wp-content/ from the Global Refugee Forum and the Impact uploads/2022/03/Chad_Covid_Refugee_ of COVID-19.” Save the Children, London. Round01_Final.pdf. WFP (World Food Programme). 2021. “Refugees ———. 2021c. “Monitoring Social and Economic Worldwide Face Rising Hunger Due to Funding Impacts of COVID-19 on Refugees in Uganda: Gaps amidst Covid-19.” News Release, June Results from the High-Frequency Phone Survey.” 8, WFP, Rome. https://www.wfp.org/news/ Second Round. World Bank, Washington, DC. refugees-worldwide-face-rising-hunger-due- https://openknowledge.worldbank.org/server/ funding-gaps-amidst-covid-19. api/core/bitstreams/9fa737e2-5b0b-5cfd-ac33- WFP, and IOM (International Organization for b87ecb7d0603/content. Migration). 2020. “Populations at Risk: ———. 2022. Poverty and Shared Prosperity 2022: Implications of COVID-19 for Hunger, Correcting Course. Washington, DC: World Migration and Displacement.” An Analysis Bank. https://openknowledge.worldbank.org/ of Food Security Trends in Major Migration server/api/core/bitstreams/b96b361a-a806- Hotspots, WFP, Rome. https://docs.wfp.org/api/ 5567-8e8a-b14392e11fa0/content. documents/WFP-0000120687/download/?_ ———. 2023a. “Do Legal Restrictions Affect ga=2.232496671.1833407868.1683068869- Refugees’ Labor Market and Education 2114909262.1683068869. Outcomes? Evidence from Harmonized Data.” World Bank. 2019a. “Informing Durable Solutions Brief 2. Leveraging Harmonized Data to for Internal Displacement in Nigeria, Somalia, Improve Welfare among Forcibly Displaced South Sudan, and Sudan. Volume A: Overview.” Populations and their Hosts, World Bank World Bank, Washington, DC. Poverty and Equity Global Practice, World Bank, ———. 2019b. “Informing the Refugee Policy Washington, DC. https://documents1.worldbank. Response in Uganda: Results from the Uganda org/curated/en/099062123080037343/pdf/ Refugee and Host Communities 2018 Household P1780290fb9f100f20994e093835a004429.pdf. Survey.” World Bank, Washington, DC. ———. 2023b. “A Profile of Forcibly Displaced ———. 2020a. “Challenges and Opportunities of Populations and their Hosts.” Brief 1. Leveraging Venezuelan Migration in Ecuador.” World Bank, Harmonized Data to Improve Welfare Washington, DC. among Forcibly Displaced Populations and ———. 2020b. “COVID-19 Impact Monitoring: their Hosts, World Bank Poverty and Equity Uganda Round 3.” World Bank, Washington, DC. Global Practice, World Bank, Washington, http://hdl.handle.net/10986/34993. DC. https://documents1.worldbank.org/ ———. 2020c. “COVID-19 to Plunge Global Economy curated/en/099062123080014273/pdf/ into Worst Recession since World War II.” Press P17802900b3a9b0060abd3096c105b8f418. Release, June 8. https://www.worldbank.org/en/ pdf. news/press-release/2020/06/08/covid-19-to- ———. 2023c. “Welfare in Forcibly Displaced plunge-global-economy-into-worst-recession- Populations: From Measuring Outcomes to since-world-war-ii. Building Capabilities.” Brief 3. Leveraging ———. 2020d. “Impacts of COVID-19 on Work and Harmonized Data to Improve Welfare Wages in Cox’s Bazar: Part 1 - Rohingya Camps.” among Forcibly Displaced Populations and Cox’s Bazar Panel Survey: Rapid Follow-Up their Hosts, World Bank Poverty and Equity Round 2, World Bank, Washington, DC. http:// Global Practice, World Bank, Washington, hdl.handle.net/10986/35674. DC. https://documents1.worldbank.org/ ———. 2020e. “Socioeconomic Impacts of curated/en/099062123080028808/pdf/ COVID-19 in Kenya on Households. Rapid P17802909b26930f30ac5c0c6a4cb117236.pdf. Response Phone Survey Round 1, January 2021.” ———. 2023d. World Development Report 2023. World Bank, Washington, DC. https://documents. Migrants, Refugees, and Societies. Washington, worldbank.org/en/publication/documents- DC: World Bank. reports/documentdetail/384651613652984513/ World Bank, and UNHCR (United Nations High socio-economic-impacts-of-covid-19-in-kenya- Commissioner for Refugees). 2021. “Global Cost results-update. Of Inclusive Refugee Education.” World Bank ———. 2021a. “COVID-19 Impact Monitoring: and UNHCR, Washington, DC and Geneva. Uganda Round 4-5.” World Bank, Washington, Zetter, Roger, and Héloïse Ruaudel. 2016. DC. http://hdl.handle.net/10986/35713. “Refugees’ Right to Work and Access to Labor ———. 2021b. “COVID-19 Impact Monitoring on Markets – An Assessment.” Part I: Synthesis. 95 Global Knowledge Partnership on Migration and Development (KNOMAD), Washington, DC. https://www.knomad.org/sites/default/ files/2017-03/KNOMAD%20Study%201%20 Part%20I-%20Assessing%20Refugees%27%20 Rights%20to%20Work_final.pdf. 96