Policy Research Working Paper 9327 The Lives and Livelihoods of Syrian Refugees in the Middle East Evidence from the 2015‒16 Surveys of Syrian Refugees and Host Communities in Jordan, Lebanon, and Kurdistan, Iraq Nandini Krishnan Flavio Russo Riva Dhiraj Sharma Tara Vishwanath Poverty and Equity Global Practice July 2020 Policy Research Working Paper 9327 Abstract The Syrian crisis has led to rapid and large-scale population conditions of refugees after displacement, documenting displacement. This paper has two main aims. (i) It docu- vulnerability along several dimensions, such as housing ments the size and timing of the Syrian refugee influx into access and quality, labor market attachment, and finan- Jordan, Lebanon, and Kurdistan, characterizing the forced cial security. The data sources include the United Nations nature of displacement and exploring factors that influ- High Commissioner for Refugees’ registration database and enced the decision to flee and subsequently move within multi-country, multi-topic surveys conducted in 2015‒16. the host country. (ii) The paper describes the daily living This paper is a product of the Poverty and Equity Global Practice. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at nkrishnan@worldbank.org, flaviorussoriva@gmail.com, dsharma5@worldbank.org, and tvishwanath@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team The Lives and Livelihoods of Syrian Refugees in the Middle East: Evidence from the 2015-16 Surveys of Syrian Refugees and Host Communities in Jordan, Lebanon, and Kurdistan, Iraq Nandini Krishnan 1, Flavio Russo Riva 2, Dhiraj Sharma1 and Tara Vishwanath1 3 JEL Classification: F22, O15 Keywords: Syrian refugees; Jordan; Lebanon; Iraq; Livelihoods 1 Poverty and Equity Global Practice, World Bank. 2 Fundação Getulio Vargas 3 Supported by Monica Biradavolu, Kevin Carey, Thomas Ginn, Jacob Goldston, Matthew Groh, Meghan Lucas, Daniel Masterson, Anna Newby, and Nethra Palaniswamy. The findings, interpretations and conclusions expressed herein are those of the authors and do not necessarily reflect the view of the World Bank Group, its Board of Directors or the governments they represent. 1. Introduction The Syrian crisis has caused one of the largest episodes of forced displacement since World War II and some of the densest refugee-hosting situations in modern history. The Syrian Arab Republic’s immediate neighbors host the bulk of Syrian refugees, with Turkey, Lebanon, and Jordan all ranking in the top five countries globally for the number of refugees hosted. In 2018, according to the United Nations High Commissioner for Refugees (UNHCR), Turkey hosted 3.5 million Syrian refugees, Lebanon 0.94 million, and Jordan 0.68 million. 4 This paper characterizes the displacement and welfare of refugees from Syria, who were living in Jordan (Amman governorate, Za’atari and Azraq camps, and areas surrounding these camps in Mafraq and Zarqa governorates), the Kurdistan Region of Iraq (KRI), and Lebanon. The analysis is based on refugee registration data from UNHCR, quantitative data from the 2015–16 Syrian Refugees and Host Communities Surveys (SRHCS), and qualitative data from focus groups and in-depth interviews. 5 The SRHCS was the first attempt at a comparable and comparative analysis of forced displacement across countries with differing policy regimes governing refugees and internally displaced people. The survey’s goal was to document host countries’ main concerns about hosting refugees and the different strategies they adopted to address the influx of forcibly displaced people. The specific objectives of the multi-country analysis were: i. To assess the socio-economic and living conditions of a representative sample of the Syrian refugee and host community populations ii. To understand the implications for social and economic conditions in host communities iii. To identify strategies to support Syrian refugees and host communities in the immediate and longer term. There were several challenges associated with conducting a representative survey of the host community population and the forcibly displaced, discussed in detail in Krishnan et al. (2018). In particular, there were no reliable sampling frames for either the resident population or the forcibly displaced. In Lebanon, survey implementation first required cartographic division of the country into geographic areas small enough for a full listing operation, which provided the sampling frame for the first stage sample. In Kurdistan, the scope of the survey was expanded to include internally displaced persons (IDPs), and a representative sample of the host community, of Syrian refugees in camps and outside camps, and of IDPs in camps and outside camps was selected. In Jordan, the original design based on a nationally representative sample proved infeasible because of inaccessibility to the new sample frame based on the 2015 Population and Housing Census. The design was then amended to include a representative sample of the Za’atari and Azraq camps, which account for 4 Lebanon and Jordan ranked first and second in the world in 2018 in number of refugees hosted relative to national population. Lebanon counted 156 refugees per 1,000 national population and Jordan 72 refugees per 1,000 residents (UNHCR, 2018). These figures are based on official UNHCR registration numbers and do not reflect the unknown number of unregistered refugees. 5 The main objective of this paper is to provide a quantitative account of Syrian refugees’ departure conditions and subsequent living conditions among host communities. Qualitative evidence is mostly summarized in footnotes. 2 the vast majority of the Syrian refugee camp population in Jordan. This was complemented by purposive samples in areas surrounding the camps in Mafraq and Zarqaa governorates and a purposive sample in the Amman governorate. Detailed information on sampling design and implementation strategy for the surveys is presented in (Aguilera, et al., 2019). The remainder of the paper is structured as follows. Section 2 characterizes the migration decision of Syrian refugees as one of forced displacement, providing evidence that there was little scope for economic decision making, in the traditional sense of the term. The largest migration flows come right after peaks in violence, as proxied by monthly death rates, and refugees report that there was little time to prepare (less than a week). Also, one observes very low absolute rates of immovable assets (such as houses) capitalization and even movable assets were destroyed at high rates before displacement. The low degree of decision-making power of Syrians under forced migration is also evident when we consider the correlates of travel time after displacement. This section also discusses refugees’ human capital profiles. Section 3 describes the daily living conditions of refugees after displacement, documenting vulnerability along several dimensions, such as housing access and quality, labor market attachment, and financial security. Section 4 concludes. 2. Characterizing Forced Displacement and the Forcibly Displaced Our analytical framework for understanding forced displacement is shaped by the rich academic literature on migration, which has evolved to incorporate violence and risk to life as a fundamental determinant. The Harris-Todaro model (Harris & Todaro, 1970) states that individuals make the choice to migrate based on the expected income differential between the origin and destination locations. Individuals move if expected income in the destination is higher than expected income at home, but not otherwise. Chin and Cortes’s updated model of migration decisions (Chin & Cortes, 2015) is based on the direct costs of migrating and expected utility from various factors in the origin and destination locations. These include economic factors, such as employment, wages, opportunities, and access to public services, and social factors such as family, friends, freedom, expected reception by the host community, safety, and risk of mortality. The expected utility of these factors is based on the best information available to individuals. Ultimately, the answer to the question of whether and when to migrate depends on the expected net benefits of migration, but danger lends an urgency to the decision because of the potentially grave consequences of staying. After an individual has removed herself from immediate danger, she is again faced with the choice to migrate. Only at this point can economic and other considerations play a role in deciding where to move. 6 Our empirical analysis of the migration decision focuses on when individuals migrate, how far they move, and what factors explain their subsequent movement after initial displacement. First, we characterize the size and timing of the influx, providing evidence that violence (as proxied by monthly death rates) precedes the decision to migrate. The timing is consistent with a context where the events forcing individuals to move were 6 It is extremely difficult to model the forced migration decision empirically, because doing so requires a sample of individuals who migrated and of those who stayed. This would require sampling migrants in host countries and individuals in the country of origin, which is not feasible due to the ongoing conflict. 3 largely unpredictable. This is also supported both by the short period of time available to prepare for migration and the limited ability to capitalize assets upon moving. Secondly, we explore correlates and descriptive statistics regarding first and subsequent destinations. In particular, we suggest that the potential for economic opportunities, financial capacity, and social networks play an important role in the decision of where to settle. We end by providing some evidence on the human capital profiles of the group of migrants that left Syria and went to neighboring countries. A. Size and Timing of the Refugee Influx The panels in Figure 1 show that the bulk of Syrian refugees registered with UNHCR in Lebanon, Jordan, and Iraq arrived before 2014 (78 percent of total refugees) and that January 2013 was the month when the highest number of Syrians crossed national borders into the neighboring Mashreq countries. Over 100,000 refugees arrived in Lebanon in the first two months of 2013, and over 50,000 arrived in Jordan in the first month of 2013. 7 8 In conflict situations, it is theoretically possible for forced displacement to precede or follow periods of violence. If individuals can anticipate spikes in violence, they may choose to leave conflict-prone regions. On the other hand, violence may destroy assets and infrastructure, and lead to deaths, injury, and loss of livelihood, forcing households to migrate in search of safety and security. Figure 2 displays the monthly death and migration rates relative to the maximum monthly flow reached in mid-2012 and early 2013, respectively. The data strongly suggest that violence preceded decisions to migrate in this setting, especially for the initial period of migration (before 2014), since the migration of Syrians out of the country is clearly associated with the level of violence experienced in the current and previous month. In particular, the peak of outmigration observed in January 2013 occurred after the Syrian civil war reached its maximum number of war deaths. The descriptive evidence is consistent with regression analysis of the timing of displacement, shown in Table 1. For every additional increase in deaths per capita in the governorate, four Syrians leave the country for Jordan in the same month and six Syrians leave in the next month. In Lebanon, the same pattern is visible for contemporary violence. However, there is no relationship between displacement and future level of violence, indicating that most of the violence that led to displacement was unpredictable from the refugees’ perspective. Remarkably, death-related variables predict roughly half of the variation of migration flows within a given month in a Syrian governorate of origin ( 2 roughly 0.5). 7 The only demographic category that changes over time is the age of refugees. Before 2014, the age distribution of refugees is very similar across the three countries. After 2014, Lebanon and Jordan restricted UNHCR from registering any new refugees. As a result, the vast majority of the new refugees registering were babies born in the host country. The age profile of registered refugees in each country is available upon request. 8 In addition to the influx of Syrian refugees, Iraq was also experiencing internal displacement due to the rise of ISIS. More than 1.3 million Iraqis were confirmed as displaced in 2014 alone (IOM-DTM) in three distinct displacement waves: (1) those displaced from the Anbar conflict starting in early January 2014; (2) the Mosul conflict, which began in early June 2014; and (3) the Sinjar crisis in which began on August 4, 2014. In total, it is estimated that 3.3 million Iraqis have been displaced since 2014. Two million of the 3.3 million estimated IDPs in Iraq reside in Kurdistan, with the peak of the IDP influx occurring between mid- 2014 and mid-2015. As of September 30, 2016, the ratio of the forcibly displaced (Syrian refugees and Iraqi IDPs) to the local KRI population was 1 to 2. 4 The forced nature of the decisions is also evident from descriptive survey data about the time available to prepare for migration and the ability to capitalize assets. Figure 3 presents the share of refugees in Kurdistan and Jordan 9 who took one day or less, between one and seven days, and more than seven days to prepare for moving. The vast majority (more than 75 percent) of refugees in Kurdistan and Jordan had a week or less to prepare, with half the Syrian refugees in KRI and 36 percent of Syrian refugees in the Jordan sample having at most a day to prepare. In addition, a natural consequence of forced displacement is the decapitalization of the fleeing population through the loss of assets. The SRHCS allows us to describe this process by comparing immovable asset ownership prior to displacement with the status of these assets upon displacement (Figure 4). Focusing on the two types of assets with higher ownership rates in 2010, dwellings and vehicles, more than 60 percent of the Syrian refugee households owned a dwelling in 2010, irrespective of their country of asylum in 2015, and more than one-third report that it was destroyed at the time of displacement. Similarly, for vehicles (cars, trucks, and motorcycles), which were owned by roughly one-fifth of refugees (results not shown). Among owners of dwellings, only a tiny fraction—ranging from less than 1 percent in Lebanon to almost 4 percent in Kurdistan—was able to sell their dwelling in preparation to leave. For vehicles, the corresponding number ranges from roughly 15 percent in Jordan to almost 50 percent for Syrian refugees in Kurdistan. B. First and Subsequent Destinations To estimate the degree of decision-making power of Syrians under forced migration, we examine the distance to their first destination of refuge. Theoretically, events and circumstances outside of people’s control largely determine where they end up under forced migration. We model how far refugees traveled by exploring the relationship between total driving time from refugees’ home districts to the districts where they initially settled in host countries, relating this to household characteristics, origin district, and timing of displacement. On average, refugees reported the following travel times to reach their destinations: 7.5 hours to Kurdistan, 4 hours to Lebanon, and 2 hours to Jordan (constant estimate in columns Jordan, Lebanon, and KRI in Table 2). The distance traveled to the first destination is determined mostly by factors outside refugees’ control: location of origin, timing of displacement, and the direct effect of the conflict on household assets and the ability to capitalize them. In Lebanon and Kurdistan, for example, refugees who left Syria in 2014 traveled significantly farther than Syrians who left in 2012. On the other hand, in Jordan, refugees traveled significantly shorter distances in years other than 2012. It is likely that host-country settlement policies arbitrarily affect the distance traveled by Syrian refugees. If all Syrians crossing into Jordan were assigned to camps after 2013, that would explain the year-specific impact of distance traveled. Home destruction is negatively related to distance traveled. But, households that were able to sell assets traveled longer distances. Likewise, refugees in Jordan and Kurdistan who had no time to prepare traveled a shorter distance. In Kurdistan, economic opportunities at the destination as measured by expected monthly income is statistically significant, but the magnitude of the influence on distance traveled is small. All in all, the evidence on the first origin is consistent with the decisions being largely determined by external 9 The question was not included in the survey in Lebanon. 5 constraints, leaving little space for optimizing behavior that characterizes economic migrants. As suggested by the regression analysis in Table 3, refugees eventually choose to go to places with higher economic opportunities, and those who have more assets are able to travel farther. Refugees with larger household size make fewer moves within the country because each movement is plausibly costlier for these households. Qualitative interviews suggest that households with social networks in the host country move significantly more often than households without social networks. 10 In Lebanon, families with higher educational attainment move more frequently than families with lower educational levels, probably because more job opportunities are available to educated individuals. C. Human Capital Profile Before considering welfare after displacement, we provide evidence on basic demographics and, especially, the education and work-experience profiles of Syrian forced migrants in neighboring countries. We first consider absolute rates of educational attainment in the UNHCR data. Then we use the data on the pre-crisis Syrian labor force and our survey to provide some evidence of selection (on observables) into migration to Lebanon, Jordan, and Kurdistan with respect to education and previous work experience. There are very few highly educated refugees in any of these countries, according to UNHCR registration data. Less than 1 percent of refugees have completed university in any of these countries, and only about 10 percent have completed high school (results not shown). The majority of refugees in each country in these data have finished between basic and middle school. Our survey suggests that the number of individuals out of school with a graduate degree (or more) might be understated in the UNHCR data, but the main conclusions about the educational profile of Syrian refugees remain qualitatively unchanged. Data from the Syrian Labor Force Survey in 2010 reveal how the average migrant compares to Syrians participating in their country’s labor market before the crisis. 10 Focus group participants in the qualitative survey corroborated the strong influence of networks in determining current location choice in Lebanon and Jordan. Refugees in Lebanon explained that their current location in the host country was largely determined by pre-existing networks: either they had a relative already living in that location who could help them, or because they (if they were male) or their husband/other male relative (if they were female) were already working in that location pre-crisis. Although social networks also mattered in Jordan, they were less driven by pre-existing work ties. Focus group participants in Jordan were just as likely to explain preference for their current location by referring to the presence of their relatives. But they were more likely than refugees in Lebanon to say that they relied on a network of friends and acquaintances in choosing their destination. Other factors cited as influencing location choice included the ability to find jobs, living among others with similar ethnicity or religious background, affordability, and perceived security and safety. Examples of accounts on the subjects collected in the focus groups were: “My cousins were living here for 15 years. I called them before coming to Lebanon so they gave me the contact of a Lebanese landlord, who told me I could build a tent on his land.” [Male, rural, Beqaa governorate, Lebanon]; “We have relatives here, my sons are here, my daughter is married in Baalbek.” [Female, urban, Beqaa governorate, Lebanon]; “I used to work here and go back every day to Syria. Then when the war started I brought my whole family here.” [Male, rural, North governorate, Lebanon]; “I am from Daraa. I came straight to Huwwarah. I entered legally. My grandfather and uncle told us to come here. We stayed with them for a few days and then found our own place.” [Male, Huwwarah, Irbid governorate, Jordan]; “I didn’t know anyone in Amman. There were some Syrians in Homs who came to Amman before me. I communicated with them and they told me that they could help find me a sponsor.” [Male, Amman city, Jordan]. 6 Figure 5 shows that the latter were almost twice as likely to have attained (or attended, in the case of Lebanon) secondary (high-school or vocational) and post-secondary education. Figure 6 presents the relative distribution of surveyed Syrian refugees across specific industries prior to the crisis, compared to the employed Syrian population. In line with the differences in educational profiles, in 2010, Syrian refugees were less likely than the typical employed Syrian to be working in high-skill jobs such as those in the public administration, health, and education industries. Workers in these jobs include, for instance, those employed in the executive and legislative administration of central, regional, and local bodies, as well as public servants, teachers and instructors, physicians, and dentists. There is also evidence that refugees were less likely to have been employed in jobs in professional services, such as financial or legal services. Finally, they were more likely have accumulated sector-specific experience in low-skill jobs in construction and “other services,” which encompasses a broad range of jobs related to the provision of services, ranging from repair and installation of equipment to transportation/storage and communication. 3. Life after Displacement This section characterizes proxies of welfare among Syrian refugees after displacement, considering shelter, labor market attachment and plausible determinants thereof, and financial security. A. Living Conditions The importance of the quality of shelter for both physical and mental health is suggested by recent research based on observational studies and randomized controlled trials. As an example of the former literature, (Cattaneo, Galiani, Gertler, Martinez, & Titiunik, 2009) used a natural experiment to show that the replacement of dirt floors with cement floors in urban slums in Mexico had a positive impact on child health and maternal mental health. Three randomized controlled trials of improvements in housing conditions provided by the TECHO NGO in Uruguay, El Salvador, and Mexico estimate a positive effect of the intervention on satisfaction with one’s own shelter and overall quality of life (Galiani, et al., 2016). They also provide evidence that child health might be affected in some settings. Our observational data for 2015 and recall data for 2010 are consistent with the possible detrimental effects associated with worse living conditions. For most refugees, forced displacement implied a transition from living in houses and apartments to living in non- standard facilities such as collective centers, worksites, and abandoned buildings, including dwellings built for purposes other than human habitation, such as garages and storage rooms. Refugees living outside camps in 2015 and 2016 – in Lebanon, this includes the entire population of refugees – have experienced an improvement in housing since that time. The share of households living in houses and apartments has increased, especially in Kurdistan and Jordan, where almost all out-of-camp refugees are now living in standard housing (Figure 7). In Lebanon, the rates upon arrival and currently show little or no improvement, roughly 75 percent and 78 percent, respectively. As a consequence of the process that led to these living conditions, refugees tend to live in crowded 7 conditions in absolute terms, 11 especially in camps. For instance, In Kurdistan and Jordan, Syrian refugees currently living in camps report occupying 0.6 rooms per capita and 0.3 bedrooms per capita (3 people per bedroom). Figure 8 characterizes the rental markets for Syrian refugees outside camps. More than 95 percent of Syrian refugees living outside camps rent their dwelling, and the vast majority report difficulty in doing so, 12 especially in Lebanon, where few landlords sign written contracts (roughly 15 percent). In the three settings considered here, rents represent a high burden in terms of total income in refugee households: at least one-third in Lebanon and significantly more than half in Jordan. 13 B. Labor Market Attachment The size and structure of the hosting economy, the right to work, and mobility are among the factors that determine the extent to which refugees are able to participate in host countries’ economies. However, the evidence from high-, middle-, and low-income countries is unequivocal in finding that refugees will find a way to work, even when they are formally prohibited from doing so. Irrespective of policies governing refugees’ rights to work, Syrian refugees in all three samples are seeking work. A little less than half of the working-age population (defined as 20 to 60-year-olds) is actively seeking work (Figure 9). However, success in finding employment has been more limited. Roughly one-third of Syrian refugees of working age in Kurdistan (or 70 percent of those in the labor force) were employed. Forty-four percent of the working-age refugee population (or almost 90 percent of those in the labor force) in Lebanon reported having worked in the last week; while in our out-of-camp Jordan sample, employment rates were the highest as a share of labor force participation at 69 percent, but still accounted for only about one-fifth of the 11 Overcrowding and the quality of housing emerged frequently in focus group discussions of how refugees are living in the host countries. For instance, in all four focus groups conducted in Kouachra, a rural site in the North governorate of Lebanon, most of the respondents said that they lived with their families in “storage rooms.” Each storage room is about the size of a one-car garage, and there are several such structures constructed one beside the other, usually on the side of the road, that were originally built for purposes other than human habitation. Some examples of statements were: “Ten of us live in one room.” [Female, Beirut]; “Rentals start from $200 for a one-room house. My whole house in narrower than this room. I live in it with my 3 children.” [Male, rural site, Mount Lebanon]; “Some of us had to live in a hen coop. We really need houses. We have no electricity and no water. The room is composed of four walls and a roof; nothing more nothing less.” [Female, Rural area, North governorate]. “I live in a one-room basement. The bathroom and kitchen are inside the room. There are bad odors, high humidity, and mold. The doctor told me while treating my son for asthma that I should move to another home. But I don’t have the money.” [Female, rural site, Beqaa governorate, Lebanon] 12 A 2014 UNHCR survey finds that rents are among the largest expenses for refugees, representing 50 percent of monthly refugee household expenditure (UNHCR, 2014). 13 Apart from camp sites, respondents across all qualitative study sites in both Lebanon and Jordan complained about being exploited by landlords who charged high rental prices for ill-maintained accommodations. They also worried about being evicted and said that they paid a large share of their earnings towards rent, sometimes forgoing food or going into debt. Some reported renting low-quality housing so that they could afford meals. First-hand accounts of the situation included the following: “When the landlord knows [that] we are Syrian, they increase the rent. The landlord tells us every time that he is spoiling us because he is only taking JD 200 as rent, as if he is renting me a castle.” [Male, Amman, Jordan]. “Housing conditions are very bad for the price. The plumbing is old, walls leak, ceilings leak, there is high humidity. If we tell the owner, he says we must make the repairs at our own expense.” [Female, Beirut, Lebanon]. 8 working-age population. An overwhelming majority of employment for refugees comes from wage work (rather than self-employment). Refugees employed in the wage sector largely do not have a written contract, except for camp refugees in Jordan who are likely to be employed by NGOs and international organizations active in the camps. Most refugees who have found work are employed in service sector jobs (Figure 10). Syrian refugees in Lebanon, despite being better educated than their counterparts in Kurdistan, work primarily in construction and household work. Roughly one-third of this group in Lebanon was employed in construction even prior to the crisis, but employment in household work (services provided to other households) has increased from 7.6 percent to 33 percent of all employment. In Kurdistan, the bulk of private sector employment for Syrian refugees comes from three industries: construction, wholesale and retail trade, and services (including household work). More than one-third of Syrian refugees were employed in construction before the crisis. While the share of refugees working in the wholesale and retail sector has increased from 6 percent to 14 percent, shares in manufacturing and agriculture sectors have fallen. In Jordan, agriculture, construction, services including household work, and wholesale and retail dominate refugees’ sectors of employment. Relatively more refugees now work in the service sector than in 2010 (20 percent compared to 7 percent). C. Financial Security A large proportion of refugees reported having faced shocks related to their incomes and cost of living over the past year. Our survey asked both refugees and host community members if they had confronted job, income, or remittance losses; faced reduced or suspended assistance; run out of savings; or experienced increases in the prices of food, housing, and other necessities. In both Lebanon and KRI, a large majority of refugees report having experienced at least one of these shocks in the past year, and at rates higher than other groups (results not shown). This feature of the lives of forcibly displaced people highlights the importance of credit markets in aiding households to cope with large insecurity. While subjective poverty rates are high for all refugees, a range of indicators also suggest improved financial well-being among refugee households in Kurdistan relative to Jordan, where similar data are available. 14 The vast majority of Syrian refugee households in Kurdistan and in the Jordan sample consider themselves to be poor and report that their earnings are inadequate to meet their basic needs. More than 60 percent report not having 100 Jordanian dinars (JDs) or 150,000 Iraqi dinars (ID) in savings (Figure 11). In fact, more than half of the refugees were in debt, owing someone outside the family at least that amount. Yet, refugees in KRI report relatively greater financial well-being than those in Jordan, including relatively lower rates of subjective poverty. In addition, refugees in KRI have better access to credit, with only one-third reporting they could not access a formal or informal loan of at least 150,000 ID from someone outside their family. In contrast, these rates are over 80 percent for a loan of 100 JDs in the Jordan sample. 14 These questions were not asked in the Lebanon survey. 9 4. Conclusion This paper sheds light on Syrian refugees’ decision to leave Syria and their living conditions once they arrived in one of the three regional host countries - Jordan, Lebanon, and the Kurdistan Region of Iraq. The analysis shows that refugees left following episodes of violence and had little time to prepare for departure. Once in the host country, they faced precarious lives, living in crowded conditions and exposed to negative shocks, with insecure employment or none at all. The data used in the analysis stem from varied sources, including UNHCR’s registration database, publicly available casualty data, and primary data collected in the three host countries. Similar efforts to collect representative data on refugees in national household surveys are underway in many countries. Such efforts are critical to generate evidence on the transition from short-term humanitarian support to a long-term development response in countries affected by protracted conflict and displacement crises. 10 References Aguilera, A., Krishnan, N., Muñoz, J., Riva, F. R., Sharma, D., & Vishwanath, T. (2019). Sampling for Representative Surveys of Displaced Populations. In J. Hoogeveen, & U. Pape, Data Collection in Fragile States: Innovations from Africa and Beyond (pp. 129-152). Washington DC: Palgrave Macmillan. Cattaneo, M. D., Galiani, S., Gertler, P. J., Martinez, S., & Titiunik, R. (2009). Housing, Health, and Happiness. American Economic Journal: Economic Policy , 75-105. Chin, A., & Cortes, K. E. (2015). The Refugee/Asylum Seeker. In B. R. Chiswick, & P. W. Miller, Handbook of the Economics of International Migration (Volume I) (pp. 585-658). Galiani, S., Gertler, P. J., Undurraga, R., Cooper, R., Martínez, S., & Ross, A. (2016). Shelter from the storm: Upgrading housing infrastructure in Latin American slums. Journal of Urban Economics, 166-194. Harris, J. R., & Todaro, M. P. (1970). Migration, Unemployment and Development: A Two-Sector Analysis. The American Economic Review, 126-142. UNHCR. (2014). Housing, Land, & Property Issues in Lebanon: Implications of the Syrian Refugee Crisis. UNHCR. (2018). Global Trends: Forced Displacement in 2018. Geneva: UNHCR. 11 Figures Figure 1. Number of Migrants (Cumulative number and Flow) Source: Authors’ calculations using registration data from UNHCR. Figure 2. Monthly Death and Migration Flows Source: Authors’ calculations using registration data from UNHCR and the deaths of combatants and non-combatants documented by Violations Documentation Center in Syria (http://www.vdc-sy.info/index.php/en/about). 12 Figure 3. Time to Prepare among Refugees 1 day or less Between 1 and 7 days Between 1 week and 1 month More than 1 month 60.0 52.1 35.9 Share (%) 40.0 27.5 23.3 23.6 16.4 13.0 20.0 8.2 0.0 Kurdistan Jordan Notes: Authors’ tabulation using data from SRHCS. The question in the survey was: “How long before you left Syria / [SOURCE GOVERNORATE] did you start preparing to move?”. The Lebanon survey did not include this question. Figure 4. Ownership, State and Capitalization Rates of Assets Owned 2010 Destroyed, if owned Sold in preparation, if owned 69.2% Lebanon 48.4% 0.8% Dwelling 60.8% Kurdistan 34.7% 3.7% 73.1% Jordan 34.8% 0.0% 18.7% Lebanon 34.2% 23.4% Vehicles 22.7% Kurdistan 18.5% 47.4% 25.5% Jordan 31.4% 16.3% Notes: Authors’ tabulation using data from SRHCS. 13 Figure 5. Education distribution: Working-age population of Syrians who became refugees in Lebanon, Jordan, and KRI (2010) and all working-age Syrians (2011) 80.0 70.0 60.0 SHARE (%) 50.0 40.0 30.0 20.0 10.0 0.0 Secondary Never Primary or Intermediat (Highschoo Graduate or attended, less e l or more illiterate Vocational) All Syrians, 2010 5.2 45.8 15.8 23.0 10.2 Lebanon, attended 13.8 31.2 37.7 14.5 2.8 Kurdistan, completed 15.0 49.6 17.1 9.8 8.5 Jordan, completed 8.1 70.5 8.3 8.4 4.8 Notes: Authors’ own tabulation using data from SRHCS. Figure 6. Distribution of the pre-crisis, working-age, employed population of Syrian refugees (2010) and all Syrians (2011) 35.6 34.2 25.5 25.0 23.0 SHARE 18.9 18.3 17.2 16.6 15.4 15.1 14.2 13.6 13.3 12.6 12.0 11.4 11.2 10.2 9.9 7.6 6.5 5.0 4.8 4.5 4.3 4.3 2.9 2.1 1.5 1.5 1.3 0.4 0.2 0.1 0.0 HOUSEHOLD WORK FOOD & BEVERAGE PR OFE SSIONAL OTHER WHOLESALE/RETAI C ONSTR UC TION M ANUFAC TUR IN G ADM INIS TR AT IO N, AGR IC ULTUR E HEALTH, AND SE R VIC E S E DUC ATION PUB LIC L INDUSTRY CATEGORY Syrian refugees, Lebanon Syrian refugees, Kurdistan Syrian refugees, Jordan All Syrians (2011) Notes: Authors’ own tabulation using data from SRHCS. 14 Figure 7. Evolution in the Share of HH living in Sub-standard Housing Upon arrival Currently 98.7% 98.3% 78.3% 78.3% 75.3% 56.5% SHARE LEBANON KURDISTAN JORDAN Notes: Authors’ own tabulation using data from SRHCS. . Figure 8. Rental Markets for Syrian Refugees outside Camps (2015) Pays rent on land or dwelling Difficulty paying rent Written rental contract Rent/HH income 99.5% 97.3% 96.8% 90.6% SHARE AMONG RENTERS 76.3% 71.4% 70.8% 69.7% 63.9% 46.9% 38.6% 14.6% LEBANON KUR DISTAN JORDAN Notes: Authors’ own tabulation using data from SRHCS. To minimize the influence of outliers and measurement error, ratio between rent and household income does not take into account households with ratio values greater than 2. Figure 9. Labor Market Outcomes, Working-Age Syrian Refugees (20-60 years old) Kurdistan Lebanon Jordan, camp Jordan, out of camp 0.97 0.94 0.93 0.88 0.81 0.79 0.78 0.71 0.69 0.66 0.56 0.50 0.48 0.47 0.45 0.32 0.29 0.08 0.05 0.01 LABOR FORCE JOBLESSNESS EMPLOYMENT WAGE CONTRACT PAR TIC IP AT IO N RATE EMPLOYMENT WORKER RATE (C ONDITION A L (C ONDITION A L (C ONDITION A L ON ON ON WAGE PAR TIC IP AT IO N) E M PLOYM E NT) EMPLOYMENT) Notes: Authors’ own tabulation using data from SRHCS. 15 Figure 10. Current Industry for Employed, Working-Age Syrian Refugees (20-60 years old) Lebanon Transportation and storage 0.66 Other services incl hh work 33.12 Public admin, health & education 2.59 Professional services 4.43 Hotels & F&B 4.76 Wholesale/retail 10.92 Construction 24 Utilities 1.55 Manufacturing 9.47 Mining and quarrying 0.17 Agriculture 8.33 Jordan Transportation and storage 1.15 Other services incl hh work 19.40 Public admin, health & education 3.00 Professional services 1.62 Hotels & F&B 3.23 Wholesale/retail 12.01 Construction 23.79 Utilities 4.39 Manufacturing 6.24 Mining and quarrying 5.31 Agriculture 19.63 Kurdistan Transportation and storage 7.17 Other services incl hh work 10.68 Public admin, health &… 4.59 Professional services 4.03 Hotels & F&B 8.59 Wholesale/retail 14.25 Construction 39.30 Utilities 1.75 Manufacturing 8.39 Mining and quarrying 0.13 Agriculture 1.13 Notes: Authors’ own tabulation using data from SRHCS. 16 Figure 11. Measures of Financial Well-Being, KRI and Jordan Kurdistan, SR (camp) Kurdistan, SR (non-camp) Jordan, SR (camp) Jordan, SR (non-camp) 0% 10% 20% 30% 40% 50% 60% 70% 80% 90%100% HH income < Minimum necessary Don't have 100 JDs (Jordan) or 150 th. IDs in savings Could not access a loan (formal or informal) of this amount Owes at least this amount Consider household to be poor Notes: Authors’ own tabulation using data from SRHCS. 17 Tables Table 1. Timing of Migration and War Casualties Dependent Variable = Number of migrants Jordan Lebanon KRI Deaths (per population) [2 month lead] 0.186 1.165 (1.333) (1.338) Deaths (per population) [1 month lead] 1.774 2.705* (1.484) (1.491) Deaths (per population) [current Month] 4.339*** 7.809*** 5.282*** (1.491) (1.342) (1.327) Deaths (per population) [1 month lag] 5.949*** 6.193*** (1.493) (1.477) Deaths (per population) [2 month lag] 1.356 1.614 (1.348) (1.334) Constant -0.001 0.000 0.000 (0.000) (0.000) (0.000) Fixed effects Governorate + Month Governorate + Month Governorate + Month Observations 784 784 784 R2 0.488 0.471 0.487 Adjusted R 2 0.466 0.449 0.466 Notes: Standard errors in parentheses; p<0.1; p<0.05; ***p<0.01 * ** 18 Table 2: Minutes Traveled to Initial Destination outside Syria and Household Characteristics Dependent variable = Minutest to Initial Destination All Jordan Lebanon KRI Kurdistan 355.973 (282.743) Lebanon 168.392 (282.421) Household Size -0.391 -0.659* -0.589 -0.331 (0.608) (0.345) (1.015) (1.448) Dependency Ratio -0.645 0.163 0.066 -2.433 (0.782) (0.445) (1.309) (1.793) Working Age Male in Household 1.343 1.708 4.087 -48.026*** (3.541) (1.736) (5.637) (13.005) Home Owner -7.351*** 1.416 -9.784*** 2.453 (2.061) (1.332) (3.482) (4.994) Home Destroyed -8.892*** -1.184 -10.443*** -1.701 (2.044) (1.369) (3.375) (5.697) Vehicle Owner -3.652 0.774 -7.988 4.309 (2.866) (1.357) (4.970) (6.319) 19 Vehicle Destroyed -5.872 -3.183 -4.569 -12.936 (4.222) (2.006) (7.139) (10.347) Vehicle Sold 3.463 -3.131 5.674 -1.704 (5.132) (3.668) (8.640) (11.035) Asset Index (Normalized) -1.232** -1.759 (0.494) (1.869) Most Assets Destroyed (Excluding House/Car) 1.438 1.429 (1.195) (4.966) Most Assets Sold (Excluding House/Car) 10.221** 5.873 (5.158) (6.154) Migrated without any Preparation Time -1.763* -15.553*** (0.999) (3.899) University (Highest Level of Education in Family) 3.48 1.352 6.229 -5.63 (4.156) (2.491) (8.465) (8.113) Highschool (Highest Level of Education in Family) 3.368 1.691 4.039 1.481 (2.562) (2.166) (4.159) (7.369) Middle School (Highest Level of Education in Family) 7.726*** 0.735 8.736** 3.718 (2.585) (2.168) (4.164) (7.396) Migrated 2011 0.633 -3.727** -0.315 11.88 20 (2.425) (1.655) (3.853) (10.771) Migrated 2013 0.044 -3.644*** -2.965 7.208 (2.251) (1.130) (3.873) (4.662) Migrated 2014 14.369*** -4.778*** 11.045** 26.622*** (2.620) (1.757) (4.425) (6.092) Migrated 2015 2.42 -5.008** 4.932 -1.459 (3.674) (2.551) (6.217) (8.177) Migrated 2016 -57.285 -4.664 -63.679 (44.713) (6.467) (73.273) Economic Opportunity 1 -178.875*** (17.990) Economic Opportunity 2 0.071** (0.030) Constant 105.848 123.465*** 273.278*** 584.077*** (281.056) (3.509) (43.318) (18.421) Fixed effects District District District District Mean 230 146 252 284 Observations 2,850 1,190 913 669 R2 0.947 0.989 0.935 0.953 Adjusted R2 0.941 0.986 0.929 0.949 Notes: Standard errors in parentheses; *p<0.1; **p<0.05; ***p<0.01 21 Table 3. Number of Moves within a Host Country after Leaving Syria and Household Characteristics Dependent Variable = Number of Moves After Initial Move All Jordan Lebanon KRI Kurdistan 0.48 (5.208) Lebanon -0.792 (5.202) Household Size -0.030*** -0.072** -0.006 -0.110*** (0.011) (0.031) (0.018) (0.029) Dependency Ratio 0.041*** 0.062 0.027 0.072** (0.014) (0.041) (0.023) (0.036) Working Age Male in Household 0.076 0.035 0.045 -0.144 (0.065) (0.158) (0.097) (0.263) Home Owner -0.002 -0.01 0.036 -0.002 (0.038) (0.121) (0.060) (0.101) Home Destroyed -0.042 0.260** -0.046 0.099 (0.038) (0.125) (0.058) (0.115) Vehicle Owner 0.177*** 0.198 0.222*** 0.037 (0.053) (0.124) (0.086) (0.128) 22 Vehicle Destroyed -0.384*** 0.107 -0.423*** 0.405* (0.078) (0.183) (0.123) (0.210) Vehicle Sold -0.054 0.327 -0.081 0.159 (0.095) (0.334) (0.149) (0.223) Asset Index (Normalized) -0.021 -0.165*** (0.045) (0.038) Most Assets Destroyed (Excluding House/Car) -0.063 -0.432*** (0.109) (0.101) Most Assets Sold (Excluding House/Car) 0.124 -0.176 (0.469) (0.125) Migrated without any Preparation Time -0.116 0.135* (0.091) (0.079) University (Highest Level of Education in Family) 0.317*** 0.328 0.650*** -0.347** (0.077) (0.227) (0.146) (0.164) High school (Highest Level of Education in Family) 0.222*** 0.078 0.170** 0.096 (0.047) (0.197) (0.072) (0.149) Middle School (Highest Level of Education in Family) 0.174*** 0.135 0.203*** -0.227 (0.048) (0.197) (0.072) (0.150) Migrated 2011 0.067 0.256* 0.048 0.723*** 23 (0.045) (0.151) (0.067) (0.218) Migrated 2013 -0.044 -0.682*** 0.008 -0.124 (0.041) (0.103) (0.067) (0.094) Migrated 2014 -0.125*** -1.031*** -0.168** -0.093 (0.048) (0.160) (0.076) (0.123) Migrated 2015 -0.342*** -1.798*** -0.238** -0.815*** (0.068) (0.232) (0.107) (0.166) Migrated 2016 -0.415 -2.296*** -0.174 (0.824) (0.589) (1.265) Economic Opportunity 1 -10.247*** (1.638) Economic Opportunity 2 0.003*** (0.001) Constant 1.783 3.721*** 0.839 2.641*** (5.177) (0.319) (0.748) (0.373) Fixed effects District District District District Mean 2 3 1 2 Observations 2,850 1,190 913 669 R2 0.316 0.323 0.171 0.387 Adjusted R2 0.248 0.201 0.105 0.339 Standard errors in parentheses; *p<0.1; **p<0.05; ***p<0.01 24