How Covid-19 financially hit urban refugees: evidence from mixed-method research with citizens and Syrian refugees in Turkey.

Peering through a lens of disasters and inequalities, this article measures the financial impacts of Covid-19 on citizens and refugee communities in Turkey during a relatively early phase of the global pandemic. Our data comes from an online survey (N = 1749) conducted simultaneously with Turkish citizens and Syrian refugees living in Turkey, followed by in-depth online interviews with Syrian refugees. Our findings indicate that the initial Covid-19 measures had a higher financial impact on Syrians than on citizens when controlled for employment, wealth, and education, among other variables. In line with the literature, our research confirms that disasters' socio-economic effects disproportionally burden minority communities. We additionally discuss how Covid-19 measures have significantly accelerated effects on refugees compared to the local population, mainly due to the structural and policy context within which forcibly displaced Syrians have been received in Turkey.

Introduction

This article assesses the Covid‐19 pandemic's immediate impacts on the socio‐economic status of both Syrian refugees and Turkish citizens within Turkey. Initial analysis suggests that migrant and minority ethnic communities have been disproportionally affected by the pandemic (Guadagno, 2020; Refugees International, 2020). Along with the health‐related impacts of Covid‐19, our findings provide answers to how disasters financially affect the most vulnerable strata of societies, including refugees. Socio‐economically disadvantaged segments of the society can barely cope with disasters, mainly due to a lack of resources and social networks (Klinenberg, 1999; Cutter et al., 2003; Reid, 2013). Building on this background, we questioned how Covid‐19, as a disaster, affected the livelihoods of Syrian refugees in Turkey when compared to locals in the first phase of the pandemic. While pundits envisaged that the post‐Covid‐19 economy would increase socio‐economic inequalities, our research reveals that disadvantaged groups in society, such as refugees, had already been more negatively affected than other socio‐economic groups, even within the first few weeks of lockdown measures. We present our findings based on an online survey conducted in April 2020, when Turkey was at a relatively early phase of the pandemic, followed by in‐depth interviews with Syrian refugees in July and August 2020.

The conceptual background of our research draws from existing studies on disasters and inequalities. In the simplest terms, disasters are natural hazards and human‐made catastrophic events causing death or damage to people and which induce ‘considerable social, political and economic disruptions’ (Smith, 2005, p. 301). Thus, social and place‐based inequalities intersect and amplify social vulnerabilities in a crisis or disaster situation (Klinenberg, 1999; Cutter et al., 2003; Reid, 2013). According to Reid (2013, p. 984), disasters create ‘a natural laboratory or experimental treatment for studying the persistence of structural social inequalities'. Hence, studying responses to disasters triggered by a combination of human action and natural hazard factors, as has been the case in the Covid‐19 pandemic, provides a fruitful lens to study the depth of existing inequalities in society.

Social class, in terms of one's access to resources, is at the centre of these inequalities, which create a ‘vulnerability to loss’ during disasters (Reid, 2013). Vulnerability to loss implies any kind of loss during disasters, including financial loss and social disruptions caused by the successful or failed redistribution of resources. Fothergill and Peek (2004) discuss how poverty places households in a very disadvantaged position at every stage of the disaster process, from preparedness to their access to recovery information. In a similar vein, Donner and Rodríguez (2008) discuss the differential effects of disasters regarding the population's changing composition. Studies have long shown that being from an ethnic and racial minority not only intersects with social class but also adds another layer of social vulnerability. In his extensive study on the 1995 Chicago Heat Wave, Klinenberg (1999) reveals the heatwave's higher disproportional impact on African‐Americans when compared to the Latino community, showing that various urban underclasses experience poverty, and therefore, disasters, differently. Following these arguments, we assess the economic impact of lockdown measures in the initial months of the Covid‐19 outbreak to understand whether being a refugee in Turkey increases the severity and impact felt from such interventions.

Our findings echo previous studies, which argued that entrenched dimensions of inequality, such as social class, nationality and race, manifest themselves at a greater level in times of disasters, such as pandemics (Klinenberg, 1999; Fothergill and Peek, 2004; Reid, 2013; Qian and Fan, 2020). In terms of the economic effects of Covid‐19 measurements on the working class population, the International Labor Organization (ILO) estimated that informal economy workers, as well as casual and temporary workers, an economic class in which refugees and migrants are overrepresented, were significantly impacted by the loss of working hours and associated income due to coronavirus measures (ILO, 2020, p. 8). Besides the vulnerability of informal economy workers during disasters, studies have also shown that larger families, especially those with elderly individuals and children, may face greater challenges than smaller families due to the struggle of providing for more people (Fussell, 2012; Reid, 2013). Furthermore, women face additional challenges during disasters, mainly due to stereotypical gender relations, family care responsibilities, or sector‐specific employment conditions (Enarson and Scanlon, 1999; Cutter et al., 2003). Finally, higher educated individuals may experience the impacts of disasters to a lesser extent, as education is highly correlated with prosperity and social welfare (Cutter et al., 2003). In addition to these layers of inequality, our analysis suggests that being a refugee in Turkey exacerbates the pandemic's economic impact in comparison to citizens when controlled for education, employment, wealth, geographical location and demographic factors.

The novelty of this research is twofold. First, despite a few studies conducted in the discipline of psychology (Özmete and Pak, 2020; Kira et al., 2021) and reports by non‐governmental organisations (NGOs) (Dempster et al., 2020), to the best of our knowledge, no research in the Turkish context has so far focused on the extent to which the economic implications of Covid‐19 measures disproportionately affect refugees. This study aims to fill this gap by comparing Syrian refugees and Turkish citizens, thus contributing to the emerging literature on the economic impacts of Covid‐19, especially in forced displacement contexts. Second, we aim to contribute to the mixed‐method research process by combining our survey results alongside in‐depth interviews with migrant and ethnic communities. Our results highlight that the context of the refugee reception in Turkey has exacerbated the financial impact of the disaster for displaced communities more greatly when compared to native citizens.

The following section provides an overview of Turkey's current situation concerning Syrian refugees and the pandemic. We then explain our online survey design and summarise the main findings of Covid‐19's initial financial impact on refugee versus native households in Turkey. Our statistical findings are substantiated with contextual information. We also use qualitative interviews to reflect on how refugees experienced and coped with the first phase of the Covid‐19 pandemic between March and August 2020. We conclude by reiterating our main findings and their broader implications on societal inequalities, as well as the limitations of our study.

Refugees and the pandemic situation in Turkey

The short‐term and mid‐term impact of Covid‐19 on the refugee population compared to citizens has yet to be studied. However, Turkey provides a suitable case for this comparison as, according to UNHCR data (2021), it is the largest refugee recipient country in the world in terms of absolute number of refugees. It currently hosts more than 3.5 million Syrian refugees registered under Temporary Protection (TP), constituting around 5 percent of Turkey's population. TP enables legal access to essential public services such as health care and education, but does not constitute fully fledged refugee status. Most Syrians in Turkey, except the 2 percent of all Syrians under TP who are living in camps, are urban refugees.

Although the government enacted new regulations in regards to work permits for Syrians under TP in January 2016, they are still difficult to access. Work permits can only be obtained if employers are willing to apply on behalf of their employees, which entails a cost and a bureaucratic process. Due to these hurdles and high social security costs, only around 66,000 Syrians living in Turkey were granted legal work permits in 2019. As a result, an overwhelming majority of the over two million Syrians of working age are excluded from the formal job market. Despite modest cash transfers to vulnerable families under the Emergency Social Safety Net (ESSN) programme funded by the EU, urban refugees’ livelihoods are mainly at the whim of market forces in the absence of state‐sponsored accommodation and access to the formal labour market.

When Turkey announced its first case of Covid‐19 on 11 March 2020, the government called on citizens to stay at home and refrain from going out unless necessary. This call was initially a recommendation rather than an enforced mandate. However, as the numbers of deaths and confirmed cases rose, the government gradually introduced rules aiming to restrict access to public areas and initiated lockdowns during evenings, weekends and public holidays. Note that the Turkish government did not implement a complete lockdown of economic production. As uttered by President Erdoğan, ‘Turkey should “keep the wheels of the economy turning”’ (Kucukgocmen et al., 2020).

Nonetheless, the slowdown of the manufacturing and construction industries, where nearly half of the refugee labour force is employed, in addition to the shutdown of the services sector, including Turkish and Syrian‐owned restaurants, cafes, bars, local stores and hairdressers, left millions of self‐employed or informally employed workers, including refugees, without any income (Relief International, 2020; Tessmann, 2020). Along with Syrian refugees, almost all segments of Turkish society, especially the urban poor, experienced a decrease in their income; some lost their jobs at the onset of the pandemic, whereas others have continued working amid declining economic revenues and the fear of contracting the disease. In May 2020, the government suspended the dismissal of employees and introduced a short‐term unemployment allowance for employees who take unpaid leave. However, this only applies to those in the formal labour market.

Closely interlinked with the world economy, the pandemic is compounding the ongoing economic downturn in Turkey. Different surveys have measured citizens’ attitudes towards the virus and the pandemic's impact on their lives (IPSOS, 2020). Humanitarian agencies have also provided a rapid assessment of the pandemic's impact on refugee communities in Turkey and elsewhere (SGDD‐ASAM, 2020; TRC and IFRC, 2020). A survey conducted by the Association for Solidarity with Asylum Seekers and Migrants (SGDD‐ASAM), a leading implementing partner of UNHCR Turkey, indicated that unemployment rates among their beneficiaries increased from 18 percent to 88 percent after the March 2020 emergence of Covid‐19 (SGDD‐ASAM, 2020, p. 14). The report also notes that 63 percent of refugees have difficulties accessing food and basic necessities (SGDD‐ASAM, 2020, p. 13).

The Minister of Health stated on 10 April that all patients living in Turkey are included in Covid‐19 statistics, irrespective of nationality or legal status. However, the Ministry did not provide further information on the nationality or legal status of confirmed patients, nor explained extra measures being taken to protect refugee communities.

Syrian refugees in Turkey, similar to other urban and camp refugees in the world, are more susceptible to the pandemic's health and economic impacts. They mostly live in overcrowded and dilapidated dwellings with other family members, households that often struggle to afford high rent prices in large, populated urban centres. These overcrowded houses not only make social distancing near impossible but also create very poor health and sanitary conditions. It is expected that refugees showing mild symptoms would refrain from visiting hospitals out of fear that they will lose their jobs and their homes, as the shared dwellings they inhabit are not suitable for self‐isolation. Additionally, in Turkey as well as elsewhere, in the absence of comprehensive paid leave for salaried workers or compensation for the self‐employed, the pandemic has caused severe financial and emotional distress among the population. Besides, the majority of Syrian refugees work in the informal labour market (Ceritoglu et al., 2017; Del Carpio and Wagner, 2015) and hence could not benefit from these formal payments.

The situation in Istanbul deserves a focus as it was the epicentre of the pandemic during the time of data collection. Istanbul is the heart of the Turkish economy, with 16 million inhabitants and over half a million Syrians registered under TP. In addition, at least 4 percent of the Syrian population are not registered or are registered under a different legal status. The metropolis of Istanbul is also a hub for migrants seeking work, particularly in the informal sector. Although Istanbul provides refugees with job opportunities, the high cost of living also creates financial burden and insecurity. Much of the urban poor, including refugees, are day labourers who work side by side in packed shopfloors and live in overcrowded households, inevitably leading to high rates of infection in Istanbul's densely populated and poor neighbourhoods. As announced by the Ministry of Health on 10 April, data on the distribution of positive Covid‐19 cases per province indicates that Istanbul ranked as the first province, encompassing nearly 60 percent of positive diagnoses across Turkey (Milliyet, 2020). Consequently, one can speculate that small enterprises, characterised by widespread informality in Istanbul, might be forced to shut down during the first phase of the pandemic and lay off their disposable, unregistered labour force, among which the uneducated urban poor and Syrian refugees are overrepresented.

Noting these conditions, we have three primary hypotheses:

1) The financial impact of lockdown measures would be more severe on the Syrian refugee population than on citizens, even after controlling for social class, geographical location and demographic factors.

2) The lower the social class, the higher the financial burden of the pandemic.

3) Respondents who reside in Istanbul are more affected by the pandemic than those living outside of Istanbul.

Data and methodology

We measured the pandemic's impact on Turkish citizens and Syrian refugees in the first three weeks after Turkey's first confirmed Covid‐19 case, conducting an online survey between 5–12 April 2020 and qualitative follow‐up interviews in July and August 2020. The novelty of our online survey is that it was conducted simultaneously with refugee and native populations in Arabic and Turkish, respectively, with the latter acting as a control group for the former. The survey period took place when lockdown measures were being gradually introduced amid spiking cases of Covid‐19. We recruited respondents through Facebook and Instagram advertisements by using the Qualtrics survey tool. On average, the survey took 32 minutes (SD = 21) for Syrians and 28 minutes (SD = 49) for Turkish citizens. In total, we had 1,749 respondents (856 Syrians and 893 Turkish citizens). Participation in the survey was voluntary and anonymous.

We did not offer any bonuses or incentives to the participants for two reasons. First, incentives may have led to measurement errors, as respondents may strive to obtain the incentive as soon as possible; in some cases, respondents may answer the questions without paying attention to the content (Elliott and Valliant, 2017). In other words, incentives reduce the goal of participation to simply a chance at winning a prize (Cobanoglu and Cobanoglu, 2003). Second, incentives may also lead to undue influence, which may affect the rule of voluntary participation in the surveys (Singer and Couper, 2008). Hence, considering all these reservations, we decided not to offer any incentives.

In using convenience sampling, we applied post‐stratification weighting for the Turkish population according to age, sex, education and NUTS‐2 regions. Although we controlled for potential biases with the Turkish sample by weighting it with population data from the Turkish Statistical Institute, we did not use weights in the Syrian sample since we did not have demographic characteristics data on the Syrian refugee population in Turkey. The official data published by the Directorate General of Migration Management (DGMM) is only limited to the age and sex of Syrians under TP and does not provide data on their level of education. Additionally, an unknown number of Syrians in Turkey are either unregistered or registered under other legal statuses.

The main issue at stake is having a convenience sample which corresponds to ‘a form of nonprobability sampling in which easily locating and recruiting participants is the primary consideration’ (Elliott and Valliant, 2017, p. 251). Most critical here is the coverage and selection bias (Valliant et al., 2018), since only respondents who have an internet‐connected smartphone, tablet, or computer and have a Facebook or Instagram account could participate in the survey. However, although we do not have data about the usage of Facebook among Syrian refugees in Turkey, previous research suggests that Facebook is widely used by this group (Ramadan, 2017).

Other types of sampling methods used to access difficult‐to‐reach populations, such as refugees, have various shortcomings. For example, random‐walk sampling in particular geographic areas restricts the sample to specific neighbourhoods, limiting its representativeness of the target population. There is also no guarantee of reaching the target population in such particular areas by random walking, which may alter the walking route and create a snowball sampling, which is another nonprobability sample. Similar to the absence of a publicly available population registry, the lack of a centralised phone directory for refugees limited researchers from conducting random‐digit‐dialing phone surveys. Phone surveys are also not particularly different from online surveys, as the sample is biased towards the respondents who have a phone and phone number (Pötzschke and Braun, 2017). On the other hand, recent studies yielded promising results on conducting online surveys with difficult‐to‐reach populations (Ersanilli and van der Gaag, 2020; Pötzschke and Weiß, 2020), which is a convenient and cost‐effective alternative. More importantly, these studies based on online surveys were successful in reaching minority groups that could not be interviewed using conventional methods (Pötzschke and Braun, 2017).

Measurement and variables

Our dependent variable is a binary variable constructed with the question, ‘Did coronavirus negatively affect you financially?’ with the answers coded as 1 (Yes) and 0 (No). We controlled for the difference between Turkish citizens and Syrians with several independent variables. We generated a dummy variable for the respondents who live in Istanbul, a city with a population of 16 million and the epicentre of the pandemic during this period. We asked respondents’ employment status and generated three dummy variables as employed, unemployed but looking for a job, and housewife.

Age and education are categorically measured. The education variable ranges from 1 (no formal education and elementary school graduates) to 4 (college and postgraduates). Similarly, categorised age groups range from 1 (18–34 years old) to 4 (55 years old and above). We asked respondents the number of children they have and generated a continuous variable ranging from 0 (no child) to 6 (6 children) for Turkish respondents and 0 (no child) to 12 (12 children) for Syrian respondents. Finally, we constructed a continuous wealth variable by asking whether the respondents have a list of appliances, taking the row means ranges from 0 to 1 (see Table 1 for descriptive statistics of variables).

Table 1

Descriptive statistics of variables. Poststratification weights are applied to the Turkish sample

	Pooled			Turkish			Syrian
Variable	N	M	SD	N	M	SD	N	M	SD
Financial situation (DV)	1,719	0.82	0.38	864	0.75	0.43	855	0.91	0.29
Age
Between 18–34	1,718	0.47	0.50	864	0.34	0.47	854	0.61	0.49
Between 35–44	1,718	0.24	0.43	864	0.23	0.42	854	0.27	0.44
Between 45–54	1,718	0.10	0.29	864	0.11	0.31	854	0.08	0.27
Above 55	1,718	0.19	0.39	864	0.32	0.47	854	0.04	0.20
Male	1,720	0.52	0.50	864	0.52	0.50	856	0.51	0.50
Education
No formal education and Elementary school	1,719	0.13	0.34	864	0.18	0.39	855	0.07	0.26
Secondary school	1,719	0.24	0.42	864	0.26	0.44	855	0.21	0.41
High school	1,719	0.30	0.46	864	0.26	0.44	855	0.35	0.48
College graduate	1,719	0.33	0.47	864	0.30	0.46	855	0.37	0.48
Number of children	1,685	1.78	1.60	841	1.51	1.23	844	2.08	1.90
Employed	1,698	0.33	0.47	852	0.39	0.49	846	0.27	0.44
Unemployed	1,698	0.25	0.43	852	0.13	0.34	846	0.38	0.49
Housewife	1,698	0.17	0.38	852	0.21	0.41	846	0.13	0.33
Wealth	1,720	0.48	0.18	864	0.56	0.17	856	0.39	0.14
Istanbul	1,711	0.23	0.42	864	0.21	0.41	847	0.24	0.43

Our preliminary analysis of the survey indicated a higher level of impoverishment for Syrian refugees in relation to their Turkish counterparts. While we use contextual factors to explain the reasons, we decided to further enrich our findings with refugees’ own narratives. As it was impossible for us to conduct face‐to‐face interviews due to health and ethical reasons (as planned in the context of our ongoing research project ADMIGOV), we decided to recruit interlocutors from our online survey. To conduct online interviews, we placed another advertisement in Arabic indicating that participants could reach the research team via e‐mail if they would like to narrate their migration experiences in more detail. In total, we received 38 e‐mails and replied to all of them, informing participants about the aims of the project, interview procedures, and ensuring the privacy of their personal data. Some participants never responded, while with others it was not possible to set a time. With the help of a research assistant who is a native Arabic speaker, we eventually set up online interviews with seven Syrian refugees living in different parts of Turkey. We provided them with an internet package to cover their data usage during the interviews to prevent them from using their own mobile data plan.

We conducted online interviews through the Zoom application with the presence of the interviewer, the interlocutor and a translator when needed. We received oral consent from refugees to audio‐record the interviews and they were given the option to turn their cameras off if they did not feel comfortable revealing themselves. However, among our interlocutors, none of them turned their camera off and all agreed to be audiotaped. All interviews were transcribed and translated into English. We asked interlocutors about their migration journey, experiences living in Turkey, and the impact of Covid‐19 measures on their lives. Although we conducted interviews with a relatively small number of refugees, our interlocutors’ narratives provided insights into the adverse effects of Covid‐19 restrictions on the livelihoods of Syrian refugees. Although we were able to reach participants from different ages, educational backgrounds and socio‐economic statuses, we could only interview men talking on behalf of their households. Rather than suggesting generalisations, we believe that their narratives complement our survey results and help us provide insights on the distinctive factors affecting the financial condition of refugees in Turkey at the onset of the pandemic.

Model selection, results, and discussion

We conducted a series of logistic regressions to test the determinants of respondents’ financial concerns during the Covid‐19 period. We added employment status variables to our analysis sequentially to estimate their effects against a relevant reference group. In total we have nine models. The first three models (Models 1–3) illustrate the results from the pooled data of Syrian and Turkish participants. Models 4–6 measure the Turkish sample and Models 7–9 the Syrian sample. To test the first hypothesis, the increased effects of being a Syrian refugee over being a Turkish citizen, we first pooled the data (See Table 2). The most striking outcome was the pandemic's perceived financial impact has been much more severe on the Syrian population than the Turkish population. The first three models suggest that Syrians have been more financially affected by the Covid‐19 measures than Turkish citizens, as the odds ratio is above one and statistically significant, even after controlling for independent variables such as social class, demographic characteristics and geographical location. Model 2's margins plot of predicted probabilities also suggests that when other independent variables hold at their means, Syrians are significantly more financially affected by Covid‐19 measures than Turkish citizens (see Figure 1) (p = 0.02). In short, results from pooled sample confirm our first hypothesis.

Table 2

Logistic regression results. Odds ratios (OR) are reported. 95 percent confidence intervals are in brackets. Poststratification weights are applied to the Turkish sample

Sample	Pooled	Pooled	Pooled	Turkish	Turkish	Turkish	Syrian	Syrian	Syrian
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8	Model 9
Age (Base: Between 18–34)
Between 35–44	1.131	1.063	1.076	1.181	1.181	1.167	0.773	0.625	0.692
	[0.750,1.706]	[0.693,1.630]	[0.701,1.651]	[0.705,1.978]	[0.694,2.010]	[0.688,1.982]	[0.403,1.485]	[0.322,1.214]	[0.361,1.326]
Between 45–54	1.083	1.057	1.027	1.015	1.012	0.975	1.281	1.154	1.105
	[0.608,1.931]	[0.580,1.926]	[0.563,1.874]	[0.518,1.991]	[0.502,2.038]	[0.482,1.970]	[0.280,5.849]	[0.273,4.874]	[0.274,4.461]
Above 55	0.786	0.848	0.806	1.002	1.004	0.929	0.435	0.480	0.487
	[0.359,1.721]	[0.387,1.857]	[0.366,1.774]	[0.402,2.500]	[0.408,2.471]	[0.373,2.313]	[0.125,1.510]	[0.123,1.875]	[0.138,1.721]
Male	1.423	1.355	1.204	1.063	1.066	0.891	3.052∗∗∗	2.221∗∗∗	2.130∗∗∗
	[0.900,2.249]	[0.874,2.099]	[0.805,1.800]	[0.577,1.957]	[0.594,1.914]	[0.518,1.533]	[1.770,5.262]	[1.294,3.813]	[1.219,3.722]
Education (Base: No formal education and elementary school
Secondary school	0.718	0.663	0.740	0.579	0.539	0.632	0.757	0.697	0.804
	[0.248,2.077]	[0.230,1.908]	[0.259,2.116]	[0.178,1.888]	[0.163,1.774]	[0.193,2.065]	[0.154,3.721]	[0.147,3.306]	[0.170,3.796]
High school	0.732	0.697	0.715	0.681	0.701	0.686	0.513	0.482	0.600
	[0.294,1.824]	[0.280,1.734]	[0.286,1.788]	[0.243,1.906]	[0.248,1.983]	[0.244,1.930]	[0.109,2.413]	[0.109,2.138]	[0.136,2.652]
College	0.487	0.437+	0.449+	0.373+	0.386+	0.362+	0.455	0.361	0.498
	[0.190,1.245]	[0.171,1.117]	[0.173,1.166]	[0.124,1.121]	[0.128,1.163]	[0.120,1.098]	[0.098,2.119]	[0.082,1.590]	[0.112,2.218]
Number of children	1.110	1.120	1.133	0.963	0.977	0.998	1.328∗	1.352∗∗∗	1.352∗∗∗
	[0.922,1.337]	[0.930,1.350]	[0.926,1.385]	[0.715,1.298]	[0.726,1.313]	[0.721,1.381]	[1.065,1.655]	[1.084,1.687]	[1.082,1.690]
Wealth	0.0661∗∗∗	0.0864∗∗∗	0.0608∗∗∗	0.104∗	0.133+	0.102∗	0.0312∗∗∗	0.0289∗∗∗	0.0170∗∗∗
	[0.012,0.363]	[0.016,0.471]	[0.011,0.330]	[0.013,0.811]	[0.016,1.098]	[0.013,0.820]	[0.004,0.272]	[0.004,0.212]	[0.002,0.126]
Istanbul	2.562∗∗∗	2.570∗∗∗	2.469∗∗∗	2.331∗∗∗	2.427∗∗∗	2.256∗	2.727∗∗∗	2.820∗∗∗	2.598∗
	[1.528,4.295]	[1.535,4.305]	[1.500,4.065]	[1.241,4.376]	[1.249,4.714]	[1.207,4.218]	[1.313,5.665]	[1.370,5.802]	[1.254,5.383]
Syrian	1.950∗∗∗	1.665∗	1.903∗∗∗
	[1.253,3.036]	[1.088,2.549]	[1.212,2.985]
Employed	0.839			1.169			0.357∗∗∗
	[0.573,1.229]			[0.742,1.840]			[0.200,0.638]
Unemployed		4.008∗∗∗			4.858∗∗∗			3.611∗∗∗
		[2.458,6.536]			[2.254,10.469]			[1.883,6.925]
Housewife			0.707			0.586			1.031
			[0.373,1.340]			[0.249,1.377]			[0.442,2.405]
N	1655	1655	1655	832	832	832	823	823	823
pseudo R2	0.126	0.149	0.127	0.073	0.095	0.077	0.152	0.164	0.129

p < 0.1

p < 0.05

p < 0.01

p < 0.001.

(Source: Authors’ calculation)

Figure 1

Margins plot of Model 2. Vertical lines indicate 95 percent confidence intervals

(Source: Authors’ calculation)

Social class, as also discussed by Reid (2013), is a common factor in explaining the financial impact of Covid‐19 on both Syrian and Turkish nationals, which confirms our second hypothesis. We measured social class in relation to wealth, employment status and the educational level of respondents. Overall, the results show that poverty or a lack of resources are significant factors creating vulnerability to loss during disasters for people from lower socio‐economic statuses (Fothergill and Peek, 2004; Donner and Rodríguez, 2008). First, as respondents’ wealth increases, the economic effects of Covid‐19 measures are less severe for both citizens and refugees alike.

Second, Syrian and Turkish participants also differ in terms of their employment status (see note 6 for employment categories). While unemployment is a significant factor increasing the negative financial impact of the pandemic, employment is a significant factor only for Syrian respondents, decreasing the adverse effects of the pandemic. We did not ask whether the respondents had lost their job during the pandemic. However, the result suggests that those respondents who were unemployed at the time of the survey had been more affected. These participants had already lost income before and during the pandemic and could not find a new job under lockdown conditions, hence they were more heavily hit by lockdown measures. On the other hand, being currently employed helped refugees alleviate the negative financial impacts of the lockdown measures, while being a currently employed citizen did not have a significant impact on reducing the financial loss during the pandemic. Finally, being a housewife did not make a difference in explaining the financial effects of the pandemic.

Third, we looked at the impact of educational attainment, which reveals differences between the Syrian and Turkish communities. College graduates were less affected than respondents without formal education or elementary school graduates in the pooled sample. However, when we separate the sample, only university graduate Turkish citizens appeared significantly less affected than those without formal education or are only primary school graduates. This finding confirms the literature on the negative relationship between education and the impacts of disasters (Cutter et al., 2003). However, unlike Turkish citizens, Covid‐19 hit Syrians regardless of their education level, which implies that higher education does not necessarily bring prosperity for refugees. Even relatively educated and highly skilled Syrians ended up in odd jobs in the Turkish labour market, a result confirmed through our qualitative data.

De‐skilling after displacement is very common among migrant communities in Turkey, including Syrian refugees (Sert, 2016). Given the current refugee reception contexts, even highly educated Syrian refugees in Turkey face difficulties in finding a job in the formal labour market with social security that matches their educational background and skills. The fact that refugees are not integrated into the labour market according to their skills explains why university education is not a significant factor in alleviating the impact of the pandemic measures for Syrian refugees, unlike their counterparts in the Turkish sample. Among our interlocutors, Lukman, who lost his job and felt the pandemic's economic impact severely during this period, is a representative example of de‐skilling. Lukman, a young man now separated from his wife, albeit not officially divorced, and living by himself in a metropolitan city in Turkey, used to be a journalist working for a news agency in Syria. He was also an activist, which inspired him to stay in Syria and help people until the war escalated in his town, at which point he moved to Turkey in 2017. ‘If you write my name in Google, you can see my essays and the news I made and most of my work,’ he tells us, signalling the importance he paid to his work as a journalist back in Syria. He came to Turkey hoping to continue his work in journalism.

Lukman thinks that as the number of refugees in the country increased, job opportunities for Syrian refugees, especially for skilled migrants, became scarce. Thus, finding a suitable job for himself was already an issue before the pandemic:

My work is far from my experiences and aspirations, so it (job opportunities) is not as I expected it to be. […] I worked in many jobs, none that were related to mine (my field). I worked in a phone shop, I worked in an accessories shop, in the transactions of legal papers for Syrians, and now I am a carpenter.

Unable to find work in his field in the Turkish labour market, Lukman ended up working as a carpenter, which he had no prior experience with, without any social security under a Turkish employer in one of the biggest cities in the country. When Covid‐19 measures were implemented, he could not find even odd jobs and became unemployed for three months. During this period, he had difficulties earning his livelihood and had to borrow money from close family and friends.

Similar to Lukman's story, Safaa's narrative is another example of how education and social status in the country of origin does not protect refugees from the harsh financial effects of the pandemic in their new country. Safaa was an engineer in Syria and his wife used to be a teacher. They decided to move to Turkey in 2017 when they could no longer survive the escalating conflict in their area. Although he was a skilled worker, his efforts to find a job suitable to his skills as an agricultural engineer failed. Like many of his Syrian counterparts, Safaa worked odd jobs in Turkey to earn his living. Safaa says that he has worked in every kind of job that one could think of. His family of three did not receive any aid, because they could not meet the criteria of the ESSN eligibility, as the latter is generally given to families with several children. Thus, Safaa's family's livelihood depended on his income from temporary jobs, from which he is paid only half the daily wage that his Turkish counterparts receive. Safaa said that while the living conditions were dire, it was still better than dying in Syria.

When Covid‐19 measurements were implemented, the factory that Safaa was working in closed and he could not find another job. Safaa says that they went through tough times, especially during Ramadan. They did not have any savings and after a month he could not pay the rent or bills. In order to pay for expenses, he had to borrow money from his brother and from a friend, who were also earning their living by working odd jobs. Despite this, it was not enough to pay the rent and the bills at the same time. Later Safaa sold some of his furniture to pay his debts. Eventually, Safaa found some daily jobs in which he was able to work once or twice a week (although this came with the risk of Covid‐19 infection), but other times he could not find any jobs for weeks. While living through these challenging times without any income and facing poverty and destitution, Safaa even thought about returning to Syria:

I thought about going back to Syria, if it is my destiny to die on a specific day […] How bad can it be? Yet, my family stopped me. If I had the money, I would go to Europe. I am forced to stay here and live this way, I have no other choice.

Only a minority of Syrian refugees in Turkey could work from home during the first months of the pandemic and enjoy the privileges of working remotely. Ali Reza is an engineer and IT specialist who has been working in the humanitarian sector since he arrived in Turkey in 2017. Enjoying the benefits of formal employment, he did not lose his job nor experience a reduction in his income, although he is worried about his future job security:

Coronavirus has a real impact on my work. For example, I have been working from home (home‐based) since almost the beginning of the pandemic in Turkey. Many employees were dismissed or fired where I work. I am personally concerned about this as I was also informed that they might dismiss me from my job too. But thankfully, I still have my job.

Ali Reza's experience is an exception rather than the rule. At the other end of the spectrum, most Syrian refugees, along with the urban poor, experienced poverty and dire conditions during this period. As in the cases of Lukman and Safaa, Ahmed was another Syrian refugee who shared how he and his family survived the first months of the pandemic. He was unemployed for four months and refused to apply for refugee aid programmes, saying he ‘did not need anyone's charity.’ Without charitable aid and their savings reduced to nothing in a month, Ahmed was sent money from his uncle and brother, who had resettled in the US, so the family could survive. Ahmad was not able to pay rent or cover any household expenses without the help of his family abroad. He says that he was lucky that he did not fall into debt, unlike many of his friends who spent all their savings, fell into debt and lost their jobs.

As stated, the financial impact of the pandemic has been more severe for the Syrian population in Turkey than for Turkish citizens, especially when also considering certain demographic factors along with the contextual factors pertaining to refugee reception. For instance, as the number of children increases, Covid‐19 financially hit Syrian respondents harder than Turks. We assume that children present more of a burden and responsibility to families with lower income. In turn, families with children feel the economic impacts of Covid‐19 measures more than those without. Although studies have already shown that larger families face additional challenges when compared to smaller families (Fussell, 2012; Reid, 2013), refugee status per se is still a significant experience in explaining the adverse effects of the pandemic. As our dataset suggests, on average Syrian participants have more children than Turkish participants and this also affects the extent of economic hardship. The margins plot of Model 8 also indicates that when other independent variables hold at their means, there is a positive relationship between number of children and economic hardship. Although women might be more vulnerable during disasters (Enarson and Scanlon, 1999; Cutter et al., 2003), our findings show that men are more negatively affected financially. Syrian men, who are more likely to participate in the labour force, experienced economic hardship more than Syrian women. Note that in regards to the impacts of gender or the number of children, there is no significant difference in the Turkish sample. However, we believe that further research is needed to better understand the gender dimension of the social and psychological impacts of Covid‐19 measures.

Finally, regression results from both the pooled and separated samples reveal that respondents who reside in Istanbul have been financially more affected by Covid‐19 measures than people from other cities in Turkey, confirming our third hypothesis. As explained above, the city had been heavily hit by the pandemic in April and May 2020 and the manufacturing and service sectors shrank severely. Our in‐depth interviews also support this finding. For instance, Abdulaziz, a young Syrian man living in Istanbul with his family of three, says that refugees who are financially dependent on aid have more difficulties in Istanbul than refugees living in small cities:

I heard some Syrian families receive a monthly stipend of around 600‐700 TL in Istanbul, which cannot really help in this expensive city. Cash aid might be useful and touch the needs of refugees residing in Antakya [a region in the Southern part of Turkey bordering Syria] where you can rent a flat for 300 TL (monthly). Thus, 600 TL support can help only unemployed refugees living in such small cities to meet their basic needs and pay their rent.

Figure 2: Margins plot of Model 8. Vertical lines indicate 95 percent confidence intervals

To summarise, Covid‐19 made both Syrian and Turkish participants financially vulnerable due to the Turkish government's measures, crowded city conditions, and the decline in job opportunities. Our statistical models indicate a strong correlation between social class and the negative financial effects of Covid‐19 measures, both for Turkish citizens and Syrian refugees in Turkey. The association is even stronger for Syrians, even after controlling for a series of other independent variables. Our statistical analysis indicates that when we hold other variables at their means, Syrians appear to have been more affected than Turkish citizens. Models 6‐9 indicate that social class is a pivotal factor for analysing the financial impacts of Covid‐19 measures on Syrians but with some nuances. For Turkish citizens, being a university graduate offers an advantage under pandemic conditions, as they are already in more secure jobs and most of these ‘white collar’ jobs can be performed remotely. However, as our interlocutors highlighted in qualitative interviews, Syrian university graduates do not enjoy such an advantage, as most were already excluded from the formal labour market before the pandemic. Unemployed respondents, both Syrians and Turks, are more affected by Covid‐19 than others, illustrating the uneven effects of the measures on various groups. For Syrians, having children is a strong indicator when it comes to explaining the economic burdens of the pandemic. Finally, Syrian men felt the financial impacts of the pandemic more than Syrian women.

Conclusion

Disasters provide a powerful lens to expose pre‐disaster inequalities during the disruption of the ‘normal’ functioning of society. This article used online survey data from Syrian refugees and the native Turkish population to demonstrate that the pandemic's economic impact on individuals is not only related to social class but also to one's refugee status. Our results illustrate that those who were already working under precarious conditions without job security have been profoundly affected by the pandemic, regardless of their legal status. Lukman, one of the interlocutors, stated that ‘I can say that it was not just the Syrians who were affected by the pandemic, all the people were.’ While we agree with this general observation, we contend that being from a displaced community is not reducible to socio‐economic status.

Our findings reveal that being a refugee in Turkey accelerates and magnifies the pandemic's perceived economic impact, even after controlling for social class, demographic factors, and geographical location. The economic vulnerability of Syrian refugees in Turkey, vis‐à‐vis their Turkish counterparts, is not solely dependent on their socio‐economic status but rather on their refugee status. Despite being the very beginning of the Covid‐19 pandemic outbreak, refugees such as Ahmad, Lukman, and Safa, among others, experienced significant impoverishment when compared to Turkish citizens, who had a similar level of education. In line with previous literature, our findings indicate that pandemic measures added another layer to the already existing social vulnerability of refugees living in Turkey (Klinenberg, 1999; Cutter et al., 2003; Reid, 2013; Donner and Rodríguez, 2008). This new layer is directly linked to the context of the refugee reception in Turkey, characterised by high dependence on the market, lack of financial support, and obstacles in finding a secure job suited to one's skills in the labour force.

While signaling the unequal outcomes of the Covid‐19 pandemic, our research also addresses the underlying causes of these outcomes, as well as the implications of the pandemic for the future. As the Syrian displacement reached its 10th year, our results show that the livelihood of Syrians in Turkey is very much tied to the functioning of the labour market, characterised by widespread informality. Regardless of education level, Syrian refugees have already paid a higher economic price, even in the first few weeks after Turkey's first case was confirmed, mainly due to their already precarious position in the Turkish labour market. As the economic crisis deepens, refugees, along with other lower socio‐economic segments of society, are likely to continue to suffer and their financial losses could deteriorate their physical well‐being and mental health. The worsening economic downturn will hit the urban refugee population in Turkey, which is already just barely surviving under conditions of socio‐economic destitution. In the absence of social protection mechanisms, the downturn in economic conditions is likely to damage the already fragile co‐existence between refugee and local communities in Turkey.

Our research design is not without limitations. First, as broadly discussed before, the results are based on convenience sampling, which is not representative. The number of qualitative interviews conducted is limited and may over‐represent rather educated, better‐off individuals from the Syrian community in Turkey and those with internet access and a desire to talk. Thus, we acknowledge that we should be careful about the claims of representativeness of our qualitative interviews. However, those individuals provided insight by narrating the level of destitution they went through. Second, our dataset is only a snapshot and our results are subject to change in the future based on medical, political and economic developments. It should be highlighted here that the government's response to the pandemic has evolved alongside global developments (Bakir, 2020). Despite the rising number of positive cases in fall 2020, the Turkish government made it clear that nation‐wide economic restrictions are no longer part of planned precautions. Such a decision has potentially stymied job losses among the urban poor at the expense of an increased risk of contracting Covid‐19. Thus, our findings are limited to Covid‐19 measures in the early phases of the pandemic.

The findings merit academic attention, as we gathered quantitative and qualitative data during the peak period of the pandemic, when uncertainties and strict measures prevailed. Hence, we believe that having a control group (Turkish citizens) and the survey's timing provided a natural experiment setting, which strengthened our findings. Despite the pandemic situation, we successfully conducted a survey with Syrians refugees, a difficult‐to‐reach population for which there is no address‐based registration in Turkey. The recruitment for qualitative interviews also served as a learning experience regarding the opportunities and challenges of conducting qualitative research during times of (de facto) lockdown and under conditions jeopardising public health.

The impacts of the approaching economic downturn in the post‐Covid‐19 world on global mobility, refugee reception, and social cohesion will undoubtedly be further discussed by the pundits. More time and data are required to assess the wider economic and social impacts of the pandemic on society and to see the extent to which it will amplify existing global, national, and regional inequalities. However, it can already be claimed that the economic impacts of the pandemic are far from equal (Cutter et al., 2003; Fothergill and Peek, 2004; Qian and Fan, 2020). Lower economic classes, ethnic minorities, irregular migrants and low‐skilled workers were the first to lose their jobs in Turkey, in the US (Long and Van Dam, 2020) and elsewhere. Paradoxically, these communities are also overrepresented among essential workers such as medical staff, delivery people, agricultural workers and cashiers, among others who need to continue working for a society to function during times of lockdowns (Pew Research Center, 2020). More research on the extent to which state policies can mitigate the wider effects of disasters on populations already living in precarious conditions would also complement our analysis (Buğra et al., 2020; Kolluoğlu and Yükseker, 2020).

Authorship Statement

Authors’ names are listed by their surnames. All authors have equally contributed to the article.

Data Availability Statement

The data used in this research is part of a larger dataset collected for ADMIGOV project which received funding from the European Union's Horizon 2020 research and innovation programme under the grant agreement No 822625. Quantitative data available on request from the authors and authors elect to not share qualitative data for privacy and ethical reasons.

Acknowledgements

The authors would like to thank Anja van Heelsum, Duygu Merve Uysal, and anonymous reviewers for their constructive comments and suggestions on the earlier version of this paper. We also want to thank Alexander Burak Franklin for his meticulous copyediting service.

Appendix

We measured Turkish skills of the respondents asking them whether they can speak, write, and read in Turkish. We coded respondents who can speak, read, and write in Turkish as 1 and 0 otherwise. Measurement of remaining variables are explained above.

Table A1: Models with Turkish Speaking variable. Logistic regression results. Odds ratios (OR) are reported. 95 percent confidence intervals are in brackets.

	Syrian Model 10	Syrian Model 11	Syrian Model 12
Age (Base: Between 18–34)
Between 35–44	0.771	0.619	0.693
	[0.401,1.482]	[0.317,1.207]	[0.360,1.333]
Between 45–54	1.347	1.219	1.140
	[0.284,6.384]	[0.279,5.334]	[0.274,4.732]
Above 55	0.444	0.488	0.495
	[0.127,1.547]	[0.126,1.885]	[0.139,1.758]
Male	3.088∗∗∗	2.244∗∗∗	2.149∗∗∗
	[1.795,5.313]	[1.309,3.846]	[1.230,3.753]
Education (Base: No formal education and elementary school)
Secondary school	0.753	0.695	0.804
	[0.152,3.728]	[0.145,3.325]	[0.170,3.806]
High school	0.497	0.466	0.589
	[0.104,2.383]	[0.103,2.106]	[0.132,2.634]
College	0.438	0.345	0.488
	[0.092,2.094]	[0.076,1.559]	[0.108,2.206]
Number of children	1.336∗∗∗	1.368∗∗∗	1.357∗∗∗
	[1.074,1.663]	[1.098,1.704]	[1.089,1.691]
Wealth	0.0297∗∗∗	0.0274∗∗∗	0.0164∗∗∗
	[0.003,0.270]	[0.004,0.211]	[0.002,0.128]
Istanbul	2.758∗∗∗	2.880∗∗∗	2.612∗∗∗
	[1.337,5.688]	[1.411,5.880]	[1.268,5.380]
Turkish speaking	1.215	1.261	1.138
	[0.693,2.129]	[0.727,2.190]	[0.658,1.966]
Employed	0.352∗∗∗
	[0.198,0.628]
Unemployed		3.696∗∗∗
		[1.931,7.072]
Housewife			1.051
			[0.448,2.464]
N	823	823	823
pseudo R2	0.153	0.165	0.129

p < 0.1

p < 0.05

p < 0.01

p < 0.001.

(Source: Authors' calculation)