Policy Research Working Paper 10670 Job Finding and Separation among Syrian refugees in Jordan and Their Hosts during the COVID-19 Pandemic Sarah Wahby Ragui Assaad Middle East and North Africa Region Office of the Chief Economist January 2024 Policy Research Working Paper 10670 Abstract Refugees face important barriers to participation in the findings show the change in these rates over time for Syrians formal market, which locks them in informal employ- to be similar to those of their Jordanian hosts prior to the ment and makes them more vulnerable to shocks. Using pandemic, with a significant divergence after the start of the data from Jordan, this paper compares the job finding and pandemic. Distinguishing between Syrians living in camps separation rates of Syrian refugees to those of their hosts and those living in host communities shows that the Syrian before and after the onset of the COVID-19 pandemic. The disadvantage was entirely explained by living in camps. This paper is a product of the Office of the Chief Economist, Middle East and North Africa Region. 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 wahby001@umn.edu and assaad@umn.edu 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 Job Finding and Separation among Syrian refugees in Jordan and Their Hosts during the COVID-19 Pandemic 1 Sarah Wahby 2 and Ragui Assaad 3 JEL codes: J15, J61 Keywords: refugees, job separation, job finding, camps, COVID-19 Topics: Refugees and Jobs, Labor Markets, Labor and Employment Laws and Regulations, Coronavirus (COVID-19) Acknowledgements This work was supported by the MENA Chief Economist Office under the labor and gender research programs (TTLs: Nelly Elmallakh and Nazmul Chaudhury). We appreciate the comments of participants in the MNACE authors’ workshop, particularly our discussant Hai-Anh Dang. 1 This paper is a product of the Office of the Chief Economist, Middle East and North Africa region. 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. 2 Corresponding author. Humphrey School of Public Affairs, University of Minnesota. Email: wahby001@umn.edu 3 Humphrey School of Public Affairs, University of Minnesota. Email: assaad@umn.edu Introduction Refugees, as economic actors, are not inherently different from members of host communities. Nevertheless, they face a different set of constraints and opportunities in economic participation (Betts et al., 2017b). Although socio-economic rights are officially granted to refugees by the 1951 Convention Relating to Refugees, they are often restricted by host countries. Host governments of developing countries, which host 86 percent of refugees (UNHCR, 2021a), are typically reluctant to grant such rights in fear of increased competition over limited resources with their own citizens. The right to work is one prominent case of conflict between refugee rights and the practices of host governments. While this right is officially recognized in the 1951 convention, regulations in host countries range from denying the right to work altogether to restricting eligibility at varying levels (Zetter & Ruaudel, 2018). Facing such barriers to entry to the formal economy along with limited humanitarian support, refugees turn to informal employment characterized by lack of job security, social insurance, and lower wages. Refugees are, therefore, particularly vulnerable to shocks (Hoseini & Dideh, 2022). We study the differential employment outcomes of refugees and the Jordanian host community before and during the COVID-19 pandemic, focusing in particular on job finding and separation rates. Syrian refugees in Jordan face substantial administrative and practical barriers to obtaining work permits. In practice, they are mostly confined to specific sectors, such as agriculture, construction and manufacturing, where employment is primarily informal (Gordon, 2019). Accordingly, over 90 percent of Syrian youth in Jordan work in regular and irregular informal jobs (Assaad et al., 2021b). Moreover, Syrian refugees in Jordan live primarily in urban host communities, with only 19 percent living in officially recognized refugee camps (UNHCR, 2022). Refugees living in host communities may experience more freedom of movement and lower 2 restrictions on job search, but they do not benefit from the same humanitarian assistance refugees in camps benefit from (Betts et al., 2017b). We investigate the determinants of the vulnerability of Syrian refugees to the COVID-19 shock in Jordan. In particular, we examine whether Syrian disadvantage during the shock is completely explained by institutional barriers reflected in precarious jobs and residency in camps, or if they experience added vulnerability by virtue of their refugee status. We use retrospective data on the job histories of young Jordanian and Syrian men between the ages of 16 and 30 years from the Survey of Young People in Jordan (SYPJ) (Assaad et al., 2021a; OAMDI, 2022) to construct a synthetic semi-annual panel dataset that tracks the job finding and separation experiences of respondents. We compare the trends of job finding and separation across the two populations before and after the onset of the COVID-19 pandemic. We find that Syrians have generally experienced lower job finding rates and higher job separation rates compared to Jordanians, although the differences were generally not statistically significant prior to the pandemic. With the onset of the pandemic, Syrians became significantly disadvantaged on both measures. Controlling for the type of employment showed that workers whose last job was informal were disadvantaged on both measures compared to formal workers regardless of nationality. Controlling for the Syrian’s residency in camps, however, showed that the Syrian disadvantage was completely driven by those living in camps. In fact, holding a work permit, contrary to expectations, was correlated with significantly lower job finding rates during the pandemic particularly because camp residents are more likely to hold a work permit. The negative effect of permits disappeared when we controlled for camp residence. Moreover, controlling for informal employment, we did not find significant difference in the experiences of refugees residing in host communities and those of Jordanians. These results are in line with previous research 3 comparing Syrian refugees in Türkiye to their hosts, which showed the native-refugee gap to be much lower than previously reported for developed countries, when accounting for the fact that Syrians in Türkiye, similar to those in Jordan, are primarily employed in the informal sector (Demirci & Güray Kırdar, 2021). We found Syrians living in camps, however, to be significantly disadvantaged compared to both Jordanians and Syrians not living in camps during the pandemic. Recent papers have specifically looked at Syrian refugees’ outcomes across living arrangements in Jordan prior to the pandemic. They found Syrians outside camps to be less likely to live under the national abject poverty line, possess more assets and report higher satisfaction with access to services and with life in general (Obi, 2021). Nonetheless, using the same data, (Ginn, 2018) found that the loss of income that camp residents face is more than offset by what they save in rent. He also found no difference in growth of employment between the two populations. This paper contributes to two strands of literature. First, it contributes to the literature on the impact of the COVID-19 pandemic on the labor market outcomes of different demographic groups. Shifts in demand during the pandemic away from face-to-face services such as hospitality and retail sectors meant that the pandemic primarily affected those who cannot perform their jobs remotely (Abay et al., 2020). Evidence suggests that these are typically workers in low paying jobs (Bartik et al., 2020) who are less educated, younger, disproportionately female, and are more likely to belong to minority and immigrant groups (Angelucci et al., 2020; Borjas & Cassidy, 2020). In short, originally disadvantaged groups bore the brunt of the pandemic (Cortes & Forsythe, 2020). In developing countries, those who could perform their jobs remotely were essentially more educated workers (Gottlieb et al., 2020). In addition, social protection networks are less prevalent in these countries, exacerbating the vulnerability of workers who lost their jobs (Garrote Sanchez et al., 2021). A similar study to ours compared the outcomes of refugees in Kenya to those of 4 Kenyan nationals and found that the former group were harder hit by the pandemic and that their recovery has been slower. The gap between those two populations was not explained by demographic characteristics, residency in camps, previous employment or sector of employment and is likely the result of an “unobservable refugee factor” (Vintar et al., 2022). Evidence from the Middle East and North Africa (MENA) region showed that pre-pandemic employment was a strong predictor of workers’ ability to preserve their jobs during the pandemic and recover afterwards. In particular, workers in the formal sector, public and private, experienced greater stability compared to workers in informal employment, especially those in irregular jobs not attached to a fixed establishment (Krafft et al., 2022). Second, more broadly, this paper contributes to the nascent literature on refugee economies. This literature stresses the importance of constructing a separate theoretical framework and empirical evidence for the study of forced migration given the different choices and constraints that forced migrants face compared to other types of migrants (Betts et al., 2017a; Ruiz & Vargas-Silva, 2013; Verme, 2017; Verme & Schuettler, 2021). Evidence from this literature suggests that refugees are less likely to be employed and earn less than other migrants and hosts (Ruiz & Vargas-Silva, 2017). Nonetheless, the native-refugee gap is much smaller in developing countries compared to developed countries especially after controlling for type of employment (Demirci and Güray Kırdar 2021). The paper underlines the necessity of translating theoretical refugee rights stipulated in international conventions and treaties into practical regulations in host countries, especially the right to work. The right to work is essential for refugee self-reliance, contribution to the economy of the host country and resilience in the face of shocks. With a majority of refugees hosted in poorer countries with a limited capacity to create jobs and to provide social protection, this highlights the importance of increased cooperation and shared responsibility from the international 5 community in managing refugee crises and securing refugee rights. Similarly, while we cannot causally ascribe our results to residency in camps since residential selection into camps vs. host community is endogenous, our results for Jordan underscore the importance of revisiting camp settings and the policies governing them. In particular, the confinement of refugees to camps and their strict closures during the pandemic, while justifiable on health grounds, appear to have resulted in adverse socio-economic outcomes for camp residents. Context As of May 2022, Jordan hosted 674,000 Syrian refugees as per UNHCR registration records (UNHCR, 2022). The actual number of Syrians in Jordan could exceed 1.3 million when accounting for the unregistered and those who do not consider themselves refugees (Krafft et al., 2019). That is more than one-tenth of the Jordanian population. The bulk of Syrian refugees arrived between 2012 and 2014, with the numbers stabilizing as of mid 2016 (Alhawarin et al., 2021). In the SYPJ sample, 95% of Syrians have been in Jordan for at least 5 years. Syrians in Jordan are mainly recognizable by their different accent. Despite being the second largest per capita refugee host in the world (UNHCR, 2021b), Jordan does not have a legal framework on refugees. The country is not a signatory of the 1951 Convention Relating to Refugees and the subsequent 1967 protocol and does not have clear domestic refugee legislation. The only framework managing refugees’ issues is the memorandum of understanding the Jordanian government signed with UNHCR in 1998. Nevertheless, in line with customary law, Jordanian law explicitly prohibits refoulement 4 and allows exemptions on the requirements for entry and residency for political asylum seekers (ILO, 2015). Accordingly, Jordan 4 The forcible return of refugees or asylum seekers to a country where they are liable to be subjected to persecution. 6 has generally maintained an open border policy toward refugees reflected in its tolerance toward Syrian refugees at the onset of the Syrian crises in 2011 (Beaujouan & Rasheed, 2020). In addition to the lack of clear legislation regarding refugees, there is no clear policy concerning their right to work in the Jordanian territory. More generally the Jordanian constitution stipulates that the right to work is reserved for Jordanian citizens. In practice, the same regulations organizing the work of non-Jordanian migrants apply to refugees. A work permit approved by the ministry of labor is required. Such a permit is conditional on proving that the job in question requires experience or skills either unavailable or not sufficiently available among Jordanians. Accordingly, Syrian refugees are allowed to work in specific sectors that typically hire migrant workers, which include construction, agriculture, manufacturing, and a mix of low skill services (Gordon, 2019; ILO, 2015). In 2016, the Jordan Compact 5 relaxed many of the restrictions related to work permits, leading to a substantial increase in the number of work permits issued. Nevertheless, those were mostly issued in the construction and agriculture sectors, i.e., sectors in which Syrians were already working without permits before the Jordan Compact. This raises questions around the extent to which the Compact has resulted in job creation versus legalizing already existing work situations (Beaujouan & Rasheed, 2020). Moreover, given the predominantly informal nature of these sectors, obtaining work permits seemed to translate into the formalization of the status of the worker and not necessarily the formalization of their work (Gordon, 2019). Syrians, also, sometimes voluntary abstain from getting work permits because of fear of losing access to humanitarian assistance. Given the insecure nature of jobs available to them, they do not want to 5 An agreement with the European Union whereby the Government of Jordan received large funds and economic benefits in return for further integration of the Syrian refugees in the labor market. In the framework of the Jordan Compact the Government of Jordan pledged to provide 200,000 job opportunities for Syrian refugees. 7 run the risk of losing both their job and the humanitarian assistance they receive, and end up without any source of livelihood (Razzaz, 2017). The Jordanian economy’s capacity to create jobs has been limited. Growth has slowed down for more than a decade under the effect of external shocks beginning with the financial crises in 2008 and followed by the general instability in the region with the onset of Arab Spring uprisings in 2010, which culminated in the influx of Syrian refugees starting in 2012. Growth reached the decade’s lowest plateau of 2 percent between 2016 and 2018. In response to the stagnant growth, unemployment increased, reaching 19 percent in 2019, and labor force participation continued to fall, reaching 34.5 percent in the same year. This ranks labor market participation in Jordan among the lowest in the world (World Bank, 2020). Over the same period, the share of informal employment has increased, a reflection of the increasing share of non-national employment. Before the influx of Syrian refugees, the Jordanian labor market had a segmented structure, which for the most part allocates Jordanians to formal jobs and non-Jordanians to informal ones. In 2016, non-Jordanians made up 31 percent of employment, among whom only one-fifth were Syrians. In the same year, while 77 percent of non-Jordanians worked in informal jobs only 32 percent of Jordanians did (Assaad & Salemi, 2019). An already difficult situation became even more challenging during the COVID-19 pandemic. Among Jordanian youth 13 percent were separated from their jobs during 2020, compared to 5 percent during 2016. Perhaps even more strikingly, over 31 percent of Syrian youth experienced job separation during 2020 compared to only 2 percent in 2016 (Assaad et al., 2021b). Such important discrepancies in job separation rates between Jordanians and Syrians are probably the result of the government’s prohibition of layoffs during the pandemic, which is only pertinent to 8 formal employment. 6 In fact, among the five MENA countries with available data on labor market outcomes during the pandemic, Jordan exhibited the greatest stability in employment over time (Krafft et al., 2022). Moreover, Jordan undertook strict closure policies especially for camps whereby entering and exiting refugee camps was prohibited, including for humanitarian workers providing assistance to refugees. Data and descriptive statistics To analyze the trends of employment across Jordanians and Syrians before and during the COVID- 19 pandemic we would ideally need panel data that covers this period. Such data does not exist. The Survey of Young People in Jordan (SYPJ) 2020 (Assaad et al., 2021a; OAMDI, 2022) offers a suitable alternative. Collected between August 2020 and October 2020 7 from a nationally representative sample of Jordanian and Syrian youth between 16 and 30 years old, the survey captures the labor market experience of this population before and during the pandemic through retrospective questions about their job history since they entered the labor market. We use data on the characteristics of the individual’s current job as well as data on all their previous jobs since they entered the labor market. Individuals can report as many jobs as they took with no upper limit. The maximum number of jobs reported in our sample was seven jobs. The other source of micro data for Jordan during the pandemic is the COVID-19 MENA Monitor (CMM) data (OAMDI, 2021). This data was collected over three waves in February, June and August 2021 through a phone survey. It also has retrospective data on individuals’ employment 6 Defence order number 6 stipulated, early April, that institutions subject to the labor law must allow their employees who were dismissed or whose services were terminated since the beginning of the pandemic to return to their work < https://www.jordantimes.com/news/local/pm-issues-defence-order-no-6-stipulating-labour-rights- under-defence-law>. 7 Additional data was collected between February and March 2021 from refugee camps. 9 statuses in February 2020, right before the pandemic. We deemed the SYPJ data to be more suited to our question since it allows to look at the trend of employment outcomes before the pandemic as well as right after the shock in the summer of 2020 and therefore allows investigating the effect of the pandemic on the two populations. The CMM data would be better suited to capture the recovery from the pandemic (Krafft et al., Forthcoming). The overlapping portion of the CMM data with the SYPJ data (February 2020 to February 2021) did not allow for a comparable analysis to the one we did in this paper. The overlap is not perfect. CMM data was collected a few months after the bulk of the SYPJ data only allowing for a comparison between the first half of 2021 (CMM) to the second half of 2020 rates (SYPJ). This is not a reasonable comparison especially in a period with rapidly changing events like the pandemic time. In addition, the time intervals are different as CMM data allows for examining annual change (February 2020 to February 2021) as opposed to semi-annual, and that is only one period which does not allow to compare the trend. In our analysis we focus on young males aged 15 and above who ever worked. Since those who never worked do not have job finding and job separation outcomes we believe including them in the sample will not add much information. Moreover, given low participation rates among women for both populations, the number of women who ever worked in the SYPJ sample is only 229 (75 of whom are Syrians). It is therefore difficult to conduct the analysis by sex. Given the difference in the experiences of women and men in the labor market, we limit the sample to male respondents for whom we have information on their first job. 8 Furthermore, since our analysis spans the period from 2016 to 2020, we limit the sample to those who were 15 or above in 2016. This results in a sample of 824 young males, among whom 314 are Syrians. 8 We show the results of the full sample, men and women, in the appendix. 10 Analyzing the SYPJ sample more generally, we find that across the two populations, approximately 75 percent of young men eventually enter their first job by year 10 after exiting education or turning 15, if they exited education prior to turning 15 (Figure 1). Since the average age of exiting school in the full sample is approximately 18, and the average current age in the full sample is 21, this means that, on average, we are capturing youth experience in the labor market 3 years after exiting school. This corresponds to 60 percent of Jordanian youth (80% of youth who will eventually make it to their first job) and 50 percent of Syrian youth (two-thirds of youth who will eventually make it to their first job). Figure 1 Cumulative Hazard of entry to first job by education and nationality Source: Assaad, Krafft, Sieverding 2021 11 As Table 1 suggests, Jordanians and Syrians have significantly different characteristics. Jordanians are significantly more educated by an average of three more years of schooling. Jordanians who have ever worked are therefore, on average, older than their Syrian counterparts, with fewer years since their first job. They also have a significantly lower total number of jobs. Moreover, considering the aforementioned segmentation of the market, while 40% of Jordanian youth have informal wage work, Syrians are almost entirely in informal wage employment. Table 1: Sample characteristics of Jordanians and Syrians Jordanian Syrian Difference Age 23.99 23.65 -0.34 {3.2} {3.0} (0.25) Years of schooling 11.89 8.65 -3.24 *** {2.7} {3.2} (0.18) Years since first job 4.50 5.08 0.58 ** {3.8} {4.1} (0.24) Type of employment for last job: informal wage 0.41 0.94 0.53 *** {0.4} {0.2} (0.03) Total number of jobs taken 1.25 1.41 0.16 *** {0.7} {0.9} (0.03) N 510 314 824 Each row is a regression of the dependent variable on binary variables for each nationality. Standard deviations are reported in brackets and standard errors are reported in parentheses. Significance * .10; ** .05; *** .01. Source: Authors’ calculations using SYPJ data The SYPJ has two main limitations. First, since the sample is restricted to youth it gives no information about older working populations. Second, retrospective data on job histories 12 potentially involve long recall periods, which could result in recall bias. This would affect accuracy of information on some job characteristics. Evidence suggests that we can draw fairly accurate information on labor market dynamics from retrospective data. More elaborate information about job characteristics such as fine distinctions between different employment statuses are less reliable (Assaad et al., 2018). In our analysis, we focus on transitions in and out of different jobs, so we have reasons to believe that the data is reliable for this purpose. Recall of job start and end dates might be trickier. This is evident in the important number of respondents who could not recall the exact month or year when they started or ended a job. Since information on starting and ending dates of a job are essential for constructing the synthetic panel, we excluded from our dataset 19 individuals who could not recall the year where they started or ended any of their jobs. It is, however, easier for respondents to recall the year they started or ended a job rather than the specific month. For example, one third of the individuals who reported the starting year of their first job had a missing starting month. To avoid losing those who did not recall the specific month, we assigned a random number between 1 to 12 when the month was missing. In doing so, we considered the following: a) Respondents who ended and started two different jobs in the same year and reported the month of ending the former job are assigned a starting month for the latter job that is equal to or greater than the month when they ended their former job. b) Respondents who ended and started two different jobs in the same year and reported the month of starting the latter job are assigned an ending month for their former job that is equal to or less than the month where they started their latter job c) Since the survey ended before the end of 2020, we make sure that in 2020 the month assigned is less than or equal to the month where the individual took the survey. 13 Bearing in mind that this data is then translated to semi-annual data, i.e. the specific month is not used, among those who reported the starting month of their different jobs, the probability of starting a job in a given half-year was approximately 50 percent. We, therefore, think that a random assignment that gives equal weight to all months (and to each of the two half years) is a sensible choice. We construct a semi-annual synthetic panel. In doing so, we assume that each individual in the sample is observed over the period from the first half year of 2016 to the second half year of 2020. Individuals who entered the job market later than the first half of 2016 are considered non- employed until entry. Using the starting and ending dates for all the jobs an individual has engaged in, we can identify the individual’s job finding and job separation events. In particular, we convert the starting and ending month of a job into the half year they belong to. For example, a person who started a job in May of 2017 is coded as found a job in the first half year of 2017. In Table 2, we present the transitions by half-year, nationality of individual, their camp residency if they are Syrian and the imputation status. Overall, 27% of the transitions that occurred had an imputed month. Two things to keep in mind when considering this table. This does not show all the respondents-time periods in our sample (8,240 observations). Since imputation will only occur when there is a transition in or out of a job, we are only including in the table those transitions and not all the periods where the individual is staying in a certain job. The total number of transitions is 726 which is less than the total number of individuals in the sample. That is because for some individuals all transitions occur prior to 2016 so they are not captured in our synthetic panel. Table 2 Transitions in and out of jobs by half, nationality, camp residency and imputation status of the month where the transition occurred Syrian Half Jordanian camp=0 camp=1 14 imputed=0 imputed=1 Total imputed=0 imputed=1 Total imputed=0 imputed=1 Total 2016h1 22 16 38 8 1 9 10 11 21 2016h2 21 15 36 1 3 4 8 5 13 2017h1 23 7 30 6 4 10 8 5 13 2017h2 28 16 44 5 4 9 2 5 7 2018h1 30 17 47 7 3 10 13 7 20 2018h2 27 11 38 5 6 11 8 5 13 2019h1 39 6 45 7 4 11 10 4 14 2019h2 29 17 46 6 2 8 14 2 16 2020h1 67 7 74 24 6 30 18 2 20 2020h2 47 3 50 12 12 23 4 27 Total 333 115 448 81 33 114 114 50 164 Table 3 in the appendix, investigating the differences between individuals by their ability to recall all months of start and end of jobs they have taken, confirms that recall is systematic and associated to significantly different characteristics of respondents. Respondents who recalled the month are significantly more educated, younger and less likely to be informally employed. Systematic recall, however, does not necessarily contradict that the half year where the individual finds or loses a job is random and therefore does not invalidate our imputation strategy. In fact, systematic differences by recall ability means that not accounting for those who did not recall the month of start or end of their job will undermine the representativeness of our sample and omit important conclusions concerning specific groups in the population. Following Borjas & Cassidy (2020) we decompose the change in employment rate from one half year to another into its two main components, the job separation rate and the job finding rate. − −1 − = −1 −1 = ∙ − ∙ −1 −1 = (1 − −1 ) − −1 Where 15 : the number of persons employed at time t : the total population : the number of persons who were not employed at time t-1 but found a job by time t : the number of persons who were employed at time t-1 but lost their job by time t −1 : the number of persons who are not employed at time t-1 −1 : the employment rate at time t-1 : the fraction of persons out of work who found a job by time t i.e., the job finding rate : the fraction of employed persons who are not working by time t i.e., the job separation rate If an individual had a start date for a job in a certain half year, he is considered to have found a job in this half year. Similarly, if an individual had an end date for a job in a certain half year, he is considered to have separated from a job in this half year. Moreover, if an individual started a job and ended it within the same half year, he is considered to have both found and separated from a job in this particular half year. Using this procedure, we can estimate the rate of job finding in a certain half year, ft, and the rate of job separation in that half year, lt. Figures 2 to 4 present the descriptive statistics of the change in employment rate and its components, job separation and job finding rates, for Jordanian and Syrian males. Data in these figures is weighted to be nationally representative of Jordanian and Syrian male youth in Jordan. The data was also seasonally adjusted by subtracting from the rates for the second half-years the difference in means between the second half-year and the first half-year. Job separation rates generally coincide across nationality except that Jordanians have lower rates during the pandemic compared to Syrians. The divergence of outcomes is clearer in job finding rates where Jordanians have evidently higher rates during the pandemic. Similarly, when we condition on informal employment, we see overlapping trends for Jordanians and Syrians across 16 the period observed except that Jordanians fare much better in job finding when the pandemic hits (Figure 2). Moreover, Syrians outside camps fare clearly better than their counterparts living in camps with the divergence becoming more evident when the pandemic hits. The findings from these descriptive statistics are confirmed by our multi-variate results. While Syrians fare worse, especially in job finding, the differences are not significant until the first half of 2020 when the pandemic hits. Conditioning on informal employment shows similar patterns of disadvantage across nationality compared to formal workers. Jordanian informal workers, however, overcome this disadvantage and catch up with their formal counterparts in the second half year of 2020. Among Syrians, those who live in camps fare worse, but the differences are only substantial with the onset of the pandemic. Figure 2 Job finding rate for young men who ever worked between 2016 and 2020 by half year and nationality (Seasonally adjusted) Source: Authors’ calculations using SYPJ data 17 Figure 3 Job finding, job separation and the change in employment for young men who ever worked between 2016 and 2020 by half year and nationality conditional on informal employment in their most recent job (Seasonally adjusted) Source: Authors’ calculations using SYPJ data Figure 4 Job finding, job separation and the change in employment for Syrian young men who ever worked between 2016 and 2020 by half year and residency in camp (Seasonally adjusted) Source: Authors’ calculations using SYPJ data 18 Empirical strategy Using the synthetic semi-annual panel data we constructed, we estimate several models for each of the two outcomes: likelihood of being separated from a job from one period to another and the likelihood of finding a job. In all models we include individual and half year fixed effects. Adding individual fixed effects absorbs all time-invariant differences across individuals, including average differences between groups of workers, such as Syrians and Jordanians. We are thus measuring differences across these groups over time as they deviate from the group average. If we do not see significant differences, this does not mean that there are no differences between the two populations, rather that there are no differences in trends over time. The first model investigates the relationship between Syrian nationality and the likelihood of finding (separating from) a job in each half year from the beginning of 2016 to the end of 2020. = + + 1 ∗ + (1) Where : equal 1 if individual i finds (separates from) a job between half-year t-1 and half-year t, 0 otherwise : Syrian indicator : Half year dummy : half year fixed effects : individual fixed effects : error term In the second model we use the same specification but, in addition, we control for the type of employment of the individual’s last job, that is their current job if they are currently employed and their last job if they are currently non-employed. Workers do not transition frequently between informal and formal jobs. In our sample only 80 individuals have changed their employment type 19 at lease once since they took their first job. The formality status is therefore generally stable and equivalent to the formality of their last job. Moreover, since hardly any Syrians have formal jobs, this is equivalent to comparing each group of informal workers to formal Jordanian workers. Non- wage workers are included as a separate category in the regression, but we do not show their results given their limited sample size. = + + 1 ∗ + 2 ∗ + (2) : dummy for Syrian whose last job was private informal, 0 otherwise. : dummy for Jordanian whose last job was private informal, 0 otherwise. In the third, fourth and fifth models we limit the sample to Syrians. We use an in-camp indicator to test the effect of living in camp among Syrians (3), a permit indicator for examining the effect of holding a work permit (4), we then account for both variables in the same model (5). = + + 1 ∗ + (3) : takes 1 if Syrian living in camp, omitted category is Syrian not living in camp = + + 1 ∗ + (4) : takes 1 if Syrian holding work permit, omitted category is Syrian not holding permit = + + 1 ∗ + 2 ∗ + (5) Finally, using the full sample we include two dummy variables for being a Syrian who lives in camp or being a Syrian who does not live in camp to have a closer look at these categories in comparison to Jordanians. In this model we also control for formality of the last job. = + + 1 ∗ + 2 ∗ +3 ∗ + (6) 20 : indicator for Syrian in camp, reference category is Syrian not in camp + Jordanian : indicator for Syrian not in camp, reference category is Syrian in camp + Jordanian : type of employment, reference group: formal employment In all models we are concerned with the vector of 1 coefficients. Those represent the marginal effect of being Syrian, informal, holding a permit or living in camp in each half year. A coefficient is negative in the case of job finding and positive in the case of job separation if we expect the category in question to be experiencing increased disadvantage on the labor market in a certain half-year. Any time this coefficient has a different sign or is not significant we can refute our hypothesis of increased disadvantage for this category in this half-year. Our estimation models do not attempt to measure the effect of the pandemic on Syrian refugees or Syrians’ disadvantage compared to Jordanians. We are mainly concerned with estimating the differential trends between Syrians and Jordanians over time, including at the time of the pandemic, while accounting for the formality status of employment, and distinguishing between Syrians in camps and those in host communities. Results We present the results of the different models in Figures 5 to 10. As we mentioned earlier, these results concern men only. Results of the full sample are in the appendix. Syrians generally have lower job finding rates. The results of both populations, start to clearly diverge in 2019 where we see significantly lower job finding rates and higher job separation rates for Syrians and are further magnified during the pandemic (Figure 4). There is, however, no significant difference between Syrians’ job finding and job separation between each half-year and the next. 21 Figure 5 Results of model 1: Effect of being Syrian on the likelihood of job separation and job finding over time Conditioning on the type of employment, we find that informal workers are significantly disadvantaged compared to formal workers regardless of nationality. We do not find, however, significant differences between Jordanian and Syrian informal workers throughout the period, except in the second half of 2020. In this half-year, Jordanian informal workers are significantly less likely to separate from jobs and significantly more likely to find jobs. The general similarity in outcomes between Jordanian and Syrian informal workers and their significant difference from 22 the formal workers suggests that an important portion of Syrian disadvantage may be explained by their concentration in informal employment. Figure 6 Results of model 2: Effect of being an informal worker on the likelihood of job separation and job finding by nationality over time. Limiting the sample to Syrians, we find that Syrians in camps experience significantly lower job finding rates during the pandemic (Figure 7). Similarly, Syrians holding work permits are significantly less likely to find jobs during the pandemic (Figure 8). While counterintuitive, running the same regression conditioning on living in camp neutralizes the permit effect and shows an even larger effect of camps than in model 3 (Figure 7). This indicates that the negative effect for holding a permit we see in model 4 is entirely explained by residency in camps given that individuals in camps are more likely to hold permits. The disadvantage of Syrian refugees living 23 in camps may be attributable to the camp closure policies that were introduced after the pandemic that were designed to keep the virus out of camps. While these policies may have been justified on public health grounds, their impact on socio-economic outcomes was substantial and needs to be kept into account. Figure 7 Results of model 3: Effect of living in camps on the likelihood of job separation and job finding among Syrians over time 24 Figure 8 Results of model 4: Effect of holding a permit on the likelihood of job separation and job finding among Syrians over time 25 Figure 9 Results of model 5: Effect of holding a permit and living in a camp on the likelihood of job separation and job finding among Syrians over time Observing such strong effect of living in camp raises the question of whether the disadvantage Syrians are experiencing compared to their Jordanian counterparts is completely driven by Syrians living in camps. Running the regression again on the whole sample accounting for nationality, residency in camps and employment status of last job we find that Syrians who do not live in camps generally have very comparable outcomes to their Jordanian counterparts, once the formality of employment has been accounted for. Syrians in camps, however, are significantly disadvantaged during the pandemic compared to Jordanians or Syrians outside camps on both measures, but particularly job finding. It is important to note that the second half year of 2020 is censored for Jordanians and Syrians outside camps but not Syrians in camps given different times of data collection for these populations. More specifically, for Jordanians and Syrians outside 26 camps we observe only two to four months of the second half of 2020 depending on when the respondent took the survey, whereas for Syrians in camps we observe the full period. A plausible result of this censoring is that we are underestimating the effect for Jordanians and Syrians outside camps and therefore overestimating the gap between those populations and Syrians in camps. This, however, does not change our conclusion much since the pattern holds for the first half of 2020 where we are not concerned about censoring, at least for job finding results. If anything, Syrians in camps may have experienced a recovery in the second half that we are not capturing, this however does not negate the fact that they were affected much more substantially by the pandemic compared to other populations. Figure 10 Results model 6: Effect of Syrian living in camp on the likelihood of job separation and job finding conditioning on informality of last job 27 To account for potential biases resulting from the random assignment of the month for individuals with missing start or end of their job we ran two robustness checks for which we show the results in the appendix. We repeated the analysis limiting the sample to those who reported the month omitting all observations with imputed month. We found that the results hold for this sample as well (Figures 16-22 in the appendix). Additionally, we limited the sample to the years 2019 and 2020 given that individuals are generally more likely to report the month for jobs taken in recent years than earlier ones. While the results are attenuated compared to the main analysis, the main conclusion holds: Syrians are significantly disadvantaged compared to Jordanians during the pandemic and this disadvantage is mainly driven by Syrians residing in camps. Conclusion Syrian refugees, like other refugees, face substantial barriers to entry to the formal labor market and are therefore confined to informal employment. Even when they obtain work permits, this often merely translates into formalization of their right to work and not necessarily their employment. Concentration in informal employment means that Syrian refugees are particularly vulnerable to shocks. Moreover, Syrian refugees in camps may be experiencing an additional layer of vulnerability stemming from restrictions on movement and their inability to compete in the labor market like refugees in host communities. In this paper we tried to explore the interplay of these factors for Syrian refugees during the COVID-19 pandemic. We compared the employment outcomes of Jordanians and Syrians over a period of four years before and during the COVID-19 pandemic. This was possible by constructing a synthetic semi-annual panel using the Survey of Young People in Jordan collected in 2020 and 2021. Running a fixed effects model, we found that Syrians and Jordanians experienced similar trends in job separation and job finding rates prior to the pandemic. With the onset of the pandemic, 28 however, Syrians were found to be significantly disadvantaged on both measures. Controlling for the type of employment showed that informal workers have seen their opportunities deteriorating in job separation and job finding compared to formal workers throughout the period regardless of nationality. The only exception was the second half of 2020 when Jordanian informal workers witnessed a significant recovery. Dissecting the Syrians by their residency in camps, we find that the Syrian disadvantage observed is completely driven by Syrians in camps. These results highlight the importance of securing the right to work for refugees and dismantling barriers to formal employment to ensure their self-reliance and resilience in the face of shocks. Given the pressure on many host countries that are primarily developing countries with limited resources and social protection capacity, this calls for more coordinated efforts internationally to support refugees and alleviate the pressure on host country labor markets. Moreover, the results suggest the importance of revisiting camp settings, reconsidering camp closure policies, and providing job assistance to residents of camps. References Abay, K. A., Tafere, K., & Woldemichael, A. (2020). Winners and Losers from COVID-19: Global Evidence from Google Search. Winners and Losers from COVID-19: Global Evidence from Google Search, June. https://doi.org/10.1596/1813-9450-9268 Alhawarin, I., Assaad, R., & Elsayed, A. (2021). Migration shocks and housing: Short-run impact of the Syrian refugee crisis in Jordan. Journal of Housing Economics, 53, 101761. https://doi.org/10.1016/j.jhe.2021.101761 Angelucci, M., Angrisani, M., Bennett, D., Kapteyn, A., & Schaner, S. (2020). Remote Work and the Heterogeneous Impact of COVID-19 on Employment and Health. National Bureau of Economic Research, 34. 29 Assaad, R., Krafft, C., & Sieverding, M. (2021a). Survey of Young People in Jordan—UNICEF ACSS Dataverse V1.0 (V1.0) [dataset]. ACSS Dataverse. Assaad, R., Krafft, C., & Sieverding, M. (2021b). Youth Transitions to Adulthood in Jordan: High Aspirations, Challenging Realities (Issue November). Assaad, R., Krafft, C., & Yassin, S. (2018). Comparing retrospective and panel data collection methods to assess labor market dynamics. IZA Journal of Development & Migration. https://doi.org/10.1186/s40176-018-0125-7 Assaad, R., & Salemi, C. (2019). The Structure of Employment and Job Creation in Jordan: 2010-2016. In C. Krafft & R. Assaad (Eds.), The Jordanian Labor Market Between Fragility and Resilience. Oxford University Press. Bartik, A. W., Bertrand, M., Lin, F., Rothstein, J., & Unrath, M. (2020). Measuring the labor market at the onset of the COVID-19 Crisis. Brookings Papers on Economic Activity, Special Edition, 239–268. https://doi.org/10.1353/eca.2020.0010 Beaujouan, J., & Rasheed, A. (2020). The Syrian Refugee Policy of the Jordanian Government. In Syrian Crisis, Syrian Refugees (pp. 47–63). https://doi.org/10.1007/978-3-030-35016-1 Betts, A., Bloom, L., Kaplan, J. D., & Omata, N. (2017a). Refugee economies: Forced displacement and development (First edition). Oxford University Press. Betts, A., Bloom, L., Kaplan, J., & Omata, N. (2017b). Refugee Economies: Forced Displacement and Development (First edit). Oxford University Press. Borjas, G. J., & Cassidy, H. (2020). The Adverse Effect of the COVID-19 Labor Market Shock on Immigrant Employment. National Bureau of Economic Research, 27243. Cortes, G. M., & Forsythe, E. (2020). The Heterogeneous Labor Market Impacts of the Covid-19 Pandemic. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.3634715 30 Demirci, M., & Güray Kırdar, M. (2021). THE LABOR MARKET INTEGRATION OF SYRIAN REFUGEES IN TURKEY (14973; IZA Discussion Paper Series, Issue December). Garrote Sanchez, D., Gomez Parra, N., Ozden, C., Rijkers, B., Viollaz, M., & Winkler, H. (2021). Who on Earth Can Work from Home? World Bank Research Observer, 36(1), 67–100. https://doi.org/10.1093/wbro/lkab002 Ginn, T. D. (2018). Prison or Sanctuary? An Evaluation of Camps for Syrian Refugees. https://www.semanticscholar.org/paper/Prison-or-Sanctuary-An-Evaluation-of-Camps- for-Ginn/1d968936d17d45bb2dc5901000253ed37d6178fb Gordon, J. (2019). Refugees and decent work lessons learned from recent refugee jobs compacts. In Employment Working Paper (Issue 256). Gottlieb, C., Saltiel, F., Gottlieb, C., & Poschke, M. (2020). Working from Home in Developing Countries Working from Home in Developing Countries (13737; Discussion Paper Series, Issue 13737). Hoseini, M., & Dideh, M. J. (2022). How do shared experiences of economic shocks impact refugees and host communities? Evidence from Afghan refugees in Iran 1 (Issue Policy Research Working Paper 9915). ILO. (2015). Access to work for Syrian refugees in Jordan: A discussion paper on labour and refugee laws and policies. In Beirut: ILO Regional Office for Arab States. Krafft, C., Assaad, R., Marouani, M. A., Cheung, R., & LaPlante, A. (2022). Are labour markets in the Middle East and North Africa recovering from the COVID-19 pandemic? (ILO/ERF Working Paper). 31 Krafft, C., Assaad, R., Marouani, M. A., Cheung, R., & LaPlante, A. (Forthcoming). Are labour markets in the Middle East and North Africa recovering from the COVID-19 pandemic? IZA Journal of Development and Migration. Krafft, C., Razzaz, S., Keo, C., & Assaad, R. (2019). The Number and Characteristics of Syrians in Jordan: A Multi-source Analysis. In C. Krafft & R. Assaad (Eds.), The Jordanian Labor Market between Fragility and Resilience (pp. 173–202). Oxford University Press. OAMDI. (2021). COVID-19 MENA Monitor Household Survey (CMMHH) (Version 5.0 Jordan- CMMHH) [dataset]. http://www.erfdataportal.com/index.php/catalog. OAMDI. (2022). Survey of Young People in Jordan—UNICEF (Version 1.0) [dataset]. Egypt: Economic Research Forum (ERF). Obi, C. T. (2021). The Impact of Living Arrangements (In-Camp versus Out-of-Camp) on the Quality of Life: A Case Study of Syrian Refugees in Jordan. https://doi.org/10.1596/1813- 9450-9533 Razzaz, S. (2017). A Challenging Market Becomes More Challenging: Jordanian Workers, Migrant Workers and Refugees in the Jordanian Labor Market. Ruiz, I., & Vargas-Silva, C. (2013). The Economics of Forced Migration. Journal of Development Studies, 49(6), 772–784. https://doi.org/10.1080/00220388.2013.777707 Ruiz, I., & Vargas-Silva, C. (2017). Are Refugees’ Labour Market Outcomes Different from Those of Other Migrants? Evidence from the United Kingdom in the 2005-2007 Period: Refugees and Other Migrants Labour Market Outcomes: Evidence from the UK. Population, Space and Place, 23(6), e2049. https://doi.org/10.1002/psp.2049 UNHCR. (2021a). Global Trends: Forced Displacement in 2020. UNHCR Global Trends, 72. 32 UNHCR. (2021b). Refugees in Jordan face escalating challenges as Syria conflict marks 10-year anniversary. https://www.unhcr.org/jo/14617-refugees-in-jordan-face-escalating- challenges-as-syria-conflict-marks-10-year-anniversary.html UNHCR. (2022). Syria Regional Refugee Response. https://data2.unhcr.org/en/situations/syria Verme, P. (2017). The Economics of Forced Displacement: An Introduction. World Bank, Washington, DC. https://doi.org/10.1596/1813-9450-8038 Verme, P., & Schuettler, K. (2021). The impact of forced displacement on host communities: A review of the empirical literature in economics. Journal of Development Economics, 150. https://doi.org/10.1016/j.jdeveco.2020.102606 Vintar, M., Beltramo, T., Delius, A., Egger, D., & Pape, U. J. (2022). Impact of COVID-19 on Labor Market Outcomes of Refugees and Nationals in Kenya (9960; Policy Research Working Paper). World Bank. (2020). Jordan Economic Monitor: Weathering the storm. https://doi.org/10.1126/science.370.6514.278 Zetter, R., & Ruaudel, H. (2018). Refugees’ right to work and access to labour markets: Constraints, challenges and ways forward. Forced Migration Review, 58(June), 4–7. 33 Appendix Figure 11 Results of model 1: Effect of Syrian on the likelihood of job separation and job finding (full sample) Figure 12 Results of model 2: Effect of informal worker on the likelihood of job separation and job finding (full sample) 34 Figure 13 Results of model 3: Effect of living in camp on the likelihood of job separation and job finding among Syrians (full sample) Figure 14 Results of model 4: Effect of holding a permit on the likelihood of job separation and job finding among Syrians (full sample) 35 Figure 15 Results of model 5: Effect of holding a permit on the likelihood of job separation and job finding among Syrians conditional on living in camp (full sample) Figure 16 Results model 6: Effect of Syrian living in camp on job separation and job finding conditioning on informality of last job (full sample) 36 Figure 17 Results of model 1: Effect of Syrian on the likelihood of job separation and job finding (recall sample) Figure 18 Results of model 2: Effect of informal worker on the likelihood of job separation and job finding (recall sample) 37 Figure 19 Results of model 3: Effect of living in camp on the likelihood of job separation and job finding among Syrians (recall sample) Figure 20 Results of model 4: Effect of holding a permit on the likelihood of job separation and job finding among Syrians (recall sample) 38 Figure 21 Results of model 5: Effect of holding a permit on the likelihood of job separation and job finding among Syrians conditional on living in camp (recall sample) Figure 22 Results model 6: Effect of Syrian living in camp on job separation and job finding conditioning on informality of last job (recall sample) 39 Figure 23 Results of model 1: Effect of Syrian on the likelihood of job separation and job finding (2019-2020) Figure 24 Results of model 2: Effect of informal worker on the likelihood of job separation and job finding (2019- 2020) 40 Figure 25 Results of model 3: Effect of living in camp on the likelihood of job separation and job finding among Syrians (2019-2020) Figure 26 Results of model 4: Effect of holding a permit on the likelihood of job separation and job finding among Syrians (2019-2020) 41 Figure 27 Results of model 5: Effect of holding a permit on the likelihood of job separation and job finding among Syrians conditional on living in camp (2019-2020) Figure 28 Results model 6: Effect of Syrian living in camp on job separation and job finding conditioning on informality of last job (2019-2020) 42 Table 3 Sample characteristics of individuals who recalled month of start and end of all jobs they have taken and those who did not recall at least one by nationality Non-recall Recall Difference Jordanian Age 23.76 24.11 0.35 {3.2} {3.1} (0.30) Years of schooling 11.11 12.29 1.18 *** {2.5} {2.6} (0.24) Years since first job 5.20 4.14 -1.06 *** {3.6} {3.8} (0.35) Type of employment for last job: 0.53 0.34 -0.19 *** informal wage {0.5} {0.5} (0.05) Total number of jobs taken 1.28 1.23 -0.05 {0.7} {0.7} (0.06) N 173 335 Syrian Age 24.01 23.41 -0.6 * {3.2} {3.0} (0.35) Years of schooling 7.80 9.22 1.42 *** {3.0} {3.1} (0.35) Years since first job 5.53 4.78 -0.75 {3.9} {4.1} (0.46) Type of employment for last job: 0.95 0.92 -0.03 informal wage {0.2} {0.3} (0.03) Total number of jobs taken 1.36 1.45 0.09 {0.8} {1.0} (0.10) N 127 187 All Age 23.86 23.85 -0.01 {3.1} {3.2} (0.23) Years of schooling 9.71 11.19 1.48 *** {3.1} {3.2} (0.23) Years since first job 5.34 4.37 -0.97 *** {3.6} {3.9} (0.28) Type of employment for last job: 0.71 0.55 -0.16 *** informal wage {0.5} {0.5} (0.04) Total number of jobs taken 1.32 1.31 -0.01 {0.7} {0.7} (0.05) N 300 522 Each row is a regression of the dependent variable on binary variables for each nationality. Standard deviations are reported in brackets and standard errors are reported in parentheses. Significance * .10; ** .05; *** .01. 43 Table 4: Results model 1 and 2 Job Job Separation Job Finding Separation Job Finding (1) (2) (3) (4) 2016h2*Syrian 0.02 -0.08 ** 0.04 -0.10 ** (0.02) (0.01) (0.03) (0.05) 2017h1*Syrian 0.03 -0.06 0.03 -0.09 * (0.02) (0.01) (0.03) (0.05) 2017h2*Syrian 0.03 -0.14 *** 0.05 -0.20 *** (0.02) (0.01) (0.03) (0.06) 2018h1*Syrian 0.04 -0.07 0.04 -0.14 ** (0.02) (0.01) (0.03) (0.07) 2018h2*Syrian 0.03 -0.09 * 0.05 * -0.15 ** (0.02) (0.01) (0.03) (0.06) 2019h1*Syrian 0.05 ** -0.14 ** 0.06 * -0.32 *** (0.02) (0.01) (0.03) (0.08) 2019h2*Syrian 0.06 ** -0.17 *** 0.07 ** -0.28 *** (0.02) (0.01) (0.03) (0.09) 2020h1*Syrian 0.09 *** -0.22 *** 0.13 *** -0.36 *** (0.03) (0.01) (0.04) (0.10) 2020h2*Syrian 0.09 *** -0.33 *** 0.11 *** -0.33 *** (0.02) (0.01) (0.03) (0.12) 2016h2*Informal 0.06 * -0.06 (0.04) (0.04) 2017h1*Informal 0.00 -0.06 (0.03) (0.05) 2017h2*Informal 0.08 -0.13 ** (0.04) (0.06) 2018h1*Informal 0.05 -0.10 * (0.04) (0.06) 2018h2*Informal 0.05 -0.08 (0.04) (0.06) 2019h1*Informal 0.02 -0.31 *** (0.03) (0.07) 2019h2*Informal 0.04 -0.21 ** (0.04) (0.09) 2020h1*Informal 0.11 -0.26 ** (0.04) (0.11) 2020h2*Informal 0.04 -0.09 (0.03) (0.12) Individual fixed effects Yes Yes Yes Yes Half year fixed effects Yes Yes Yes Yes N 5361 3446 5283 3409 Note: Standard errors are clustered at the individual level and reported between parentheses. Significance * .10; ** .05; *** .01. 44 Table 5 Results model 3, 4, 5 Job Job Job Job Job Job Separation Finding Separation finding Separation finding 2016h2*Camp 0.02 0.00 0.03 -0.03 (0.05) (0.06) (0.06) (0.08) 2017h1*Camp 0.03 -0.14 * 0.03 -0.19 ** (0.04) (0.07) (0.06) (0.09) 2017h2*Camp -0.01 -0.11 -0.04 -0.13 (0.04) (0.07) (0.04) (0.09) 2018h1*Camp 0.04 -0.07 0.01 -0.12 (0.05) (0.09) (0.06) (0.12) 2018h2*Camp -0.01 -0.13 -0.04 -0.14 (0.05) (0.09) (0.05) (0.10) 2019h1*Camp 0.00 -0.13 -0.03 -0.01 (0.05) (0.10) (0.06) (0.15) 2019h2*Camp 0.05 -0.11 0.03 -0.12 (0.05) (0.10) (0.06) (0.15) 2020h1*Camp 0.06 -0.54 *** -0.05 -0.70 *** (0.06) (0.12) (0.05) (0.15) 2020h2*Camp 0.10 * -0.37 ** -0.02 -0.60 ** (0.06) (0.16) (0.05) (0.25) 2016h2*Permit -0.06 -0.04 -0.06 -0.03 (0.04) (0.06) (0.05) (0.08) 2017h1*Permit 0.01 -0.04 0.00 0.02 (0.04) (0.08) (0.04) (0.09) 2017h2*Permit -0.04 -0.10 -0.03 -0.05 (0.03) (0.07) (0.04) (0.09) 2018h1*Permit 0.04 0.05 0.04 0.10 (0.04) (0.11) (0.04) (0.13) 2018h2*Permit -0.01 -0.05 0.01 0.01 (0.04) (0.10) (0.04) (0.10) 2019h1*Permit -0.02 -0.13 -0.01 -0.13 (0.04) (0.12) (0.05) (0.14) 2019h2*Permit -0.04 -0.15 -0.05 -0.11 (0.04) (0.12) (0.05) (0.14) 2020h1*Permit -0.02 -0.27 * 0.00 0.06 (0.04) (0.15) (0.04) (0.14) 2020h2*Permit -0.02 -0.37 * -0.02 -0.16 (0.04) (0.19) (0.04) (0.22) Individual fixed effects Yes Yes Yes Yes Yes Yes Half year fixed effects Yes Yes Yes Yes Yes Yes N 2057 1277 1753 844 1753 844 Note: Standard errors are clustered at the individual level and reported between parentheses. Significance * .10; ** .05; *** .01. 45 Table 6 Results model 6 Job Job Loss finding 2016h2*Syrian in Camp -0.01 -0.05 (0.05) (0.05) 2017h1*Syrian in Camp 0.04 -0.09 * (0.04) (0.05) 2017h2*Syrian in Camp -0.02 -0.13 ** (0.05) (0.05) 2018h1*Syrian in Camp 0.03 -0.04 (0.05) (0.07) 2018h2*Syrian in Camp -0.01 -0.10 (0.05) (0.07) 2019h1*Syrian in Camp 0.04 -0.06 (0.05) (0.07) 2019h2*Syrian in Camp 0.06 -0.13 * (0.05) (0.08) 2020h1*Syrian in Camp 0.06 -0.30 *** (0.06) (0.08) 2020h2*Syrian in Camp 0.13 ** -0.37 *** (0.06) (0.10) 2016h2*Syrian not in Camp -0.02 -0.05 (0.03) (0.05) 2017h1*Syrian not in Camp 0.01 0.05 (0.02) (0.07) 2017h2*Syrian not in Camp 0.00 -0.03 (0.04) (0.06) 2018h1*Syrian not in Camp 0.00 0.02 (0.03) (0.08) 2018h2*Syrian not in Camp 0.01 0.03 (0.03) (0.08) 2019h1*Syrian not in Camp 0.04 0.06 (0.03) (0.09) 2019h2*Syrian not in Camp 0.02 -0.02 (0.03) (0.10) 2020h1*Syrian not in Camp 0.01 0.24 ** (0.04) (0.12) 46 2020h2*Syrian not in Camp 0.03 -0.01 (0.02) (0.17) 2016h2*Informal 0.06 * -0.06 (0.03) (0.04) 2017h1*Informal 0.00 -0.07 (0.03) (0.04) 2017h2*Informal 0.06 -0.12 ** (0.04) (0.05) 2018h1*Informal 0.04 -0.11 * (0.04) (0.06) 2018h2*Informal 0.05 -0.09 (0.03) (0.06) 2019h1*Informal 0.02 -0.31 *** (0.03) (0.07) 2019h2*Informal 0.04 -0.21 ** (0.03) (0.09) 2020h1*Informal 0.11 *** -0.26 ** (0.04) (0.10) 2020h2*Informal 0.04 -0.10 (0.03) (0.12) Individual fixed effects Yes Yes Half year fixed effects Yes Yes N 5283 3409 Note: Standard errors are clustered at the individual level and reported between parentheses. Significance * .10; ** .05; *** .01. 47