Who welcomes refugees to the public purse? Evidence from a survey in Turkey

Anna Getmansky, London School of Economics and Political Science, Tolga Sınmazdemir, Bogazici University, and Thomas Zeitzoff, American University[1]

*This memo is part of POMEPS Studies 31, Social Policy in the Middle East and North Africa. Download the full PDF here.


The Syrian civil war has forced more than 5.5 million people to flee their country. The great majority of those who left Syria have settled in neighboring countries. Among these neighboring countries, Turkey hosts the largest number of Syrians, with about 3.5 million Syrians currently residing there.[2] As of June 2018, there are 20 camps hosting Syrian refugees, mostly in Turkish provinces bordering Syria.[3] These camps are run and financed by the Turkish government, and they provide refugees with services in education and health. However, only a very small share (about 6 percent) of Syrians resides in these camps. The rest lives outside camps, in poor urban neighborhoods of Turkey, side by side with the local communities. Most of these out-of-camp Syrians are in need of social protection (Erdoğan, 2017).

Turkey has adopted a number of social policies to assist Syrians. Syrians who are registered with Turkish authorities have access to free primary, secondary, and tertiary healthcare in public hospitals of the provinces in which they are registered. Syrian children’s access to public education is limited but has been expanding. There are about one million school-aged Syrian children in Turkey. In the 2017 to 2018 academic year, close to 60 percent of these children have received education, with about 35 percent taught in Turkish public schools free of charge (Erdoğan, 2017). The rest received education in “temporary education centers,” whose language of instruction is Arabic. These centers are run by NGOs in collaboration with the Turkish Ministry of Education and receive their funding from a mix of organizations, including NGOs, municipal governments, and the Ministry of Education (Aras and Yasun, 2016). Turkey is in the process of phasing out these centers and integrating all Syrian children into the Turkish public education system (International Crisis Group, 2018). The financial costs of Syrians’ education for the Turkish taxpayers should therefore be expected to increase accordingly.

About a million Syrians in Turkey work in the informal sector, for wages below the minimum wage, without any social insurance (Erdoğan, 2017). In 2016, the Turkish state started to grant work permits for Syrians. Although the number of those with work permits was only about 20,000 at the end of 2017,[4] those with permits become eligible for social security benefits as the Turkish employees in the formal sector.

Recently, two new schemes of social assistance programs for Syrians have been put in effect. One is the Emergency Social Safety Net, which started in 2016 and provides cash support to eligible Syrian families. The second is the Conditional Cash Transfer for Education Scheme, which started in 2017 and provides cash assistance to Syrian families whose children attend school regularly. Although not financed by the Turkish government, these programs have exacerbated the discontent among Turkish citizens that the Syrians live on assistance by the Turkish state at the expense of locals (International Crisis Group, 2018).

While there is evidence of Turkish citizens’ discontent with public assistance to Syrians, there is very little systematic research on the determinants of this discontent. Which groups among the local Turkish population are more or less supportive of assistance to Syrian refugees? To answer this question, we conducted a survey among Turkish citizens in the summer of 2014.

Data and Research Design


Figure 1 shows the geographic distribution of our survey sample.[5] Our sample consists of respondents from 33 districts in 17 provinces.[6] The white areas represent our sampling frame, and the districts with dark borders are those where our survey took place. Our survey involved respondents from a diverse set of districts according to three factors: refugee presence, whether a district was under state of emergency rule (called OHAL by its Turkish acronym) during the clashes between Turkish state and the Kurdish insurgency PKK between 1987 to 2002, and support for the incumbent AKP political party. The list of the provinces, districts, and the number of respondents in each district is in Table 3 in the Appendix.

We surveyed 1,257 respondents.[7] Summary statistics of the variables included in the analysis are shown in Table 1. In regressions not reported here, we show that having a refugee camp in the district and being an urban resident are positively associated with the self-reported refugee exposure of our respondents in the survey. Hence, our measure of refugee exposure is valid.

Figure 1: Map of the Sampled Districts

Table 1: Summary Statistics
Variable Mean Std. Dev. Min Max N
Female 0.496 0.5 0 1 1242
Age 2.438 1.091 1 4 1242
Education 0.354 0.478 0 1 1240
Kurdish 0.424 0.494 0 1 1242
AKP supporter 0.503 0.5 0 1 1160
CHP supporter 0.134 0.34 0 1 1160
Urban area 0.731 0.444 0 1 1242
OHAL province 0.401 0.49 0 1 1242
Ramadan 0.104 0.305 0 1 1242
Components of Wealth
Smartphone 0.263 0.441 0 1 1242
Car 0.265 0.441 0 1 1242
Computer 0.4 0.49 0 1 1242
Washing Machine 0.913 0.282 0 1 1242
Dish Washer 0.564 0.496 0 1 1242
Components of Religious
Cover Hair 5.504 1.881 1 7 1220
Alcohol not OK 4.507 2.455 1 7 1233
Pray 4.372 1.927 1 7 1194
Components of self-reported Refugee Exposure
Public Transport 0.599 0.396 0 1 1211
Street 0.663 0.37 0 1 1209
Business 0.26 0.389 0 1 1205
Social Life 0.424 0.412 0 1 1197
Markets 0.593 0.388 0 1 1205
People help strangers 0.509 0.323 0 1 1235

Wealth, Religious, and Refugee Exposure are factor variables created using the components listed below each of them. The differences in the number of respondents are due to missing values.

Dependent variables

Our main dependent variables are support for public assistance to Syrian refugees in health, education, housing, and employment.[8] Figure 2 shows that more than 50 percent of our respondents supported assisting Syrian refugees in health, while close to 50 percent supported assistance in education. However, only about 35 percent supported assisting Syrians in employment and housing.

Figure 2: Responses to the Dependent Variables Questions

The plots represent the answers to our four questions on public assistance for Syrians in health, education, employment and housing. The answers range from strongly opposing assistance to strongly supporting it.


Table 3 presents the results of four OLS regressions. Respondents from former OHAL provinces that have suffered the most from the Turkish-Kurdish conflict are more supportive of public assistance to Syrians in health, education, housing, and employment. A growing body of research shows that individuals exposed to violence exhibit more cooperative and altruistic behavior, especially towards in-group members (Bauer et al., 2016). Our results show that exposure to violence can be associated with improved altruism also towards out-group members, such as refugees. These findings are also consistent with our previous research based on the same survey, which shows that Turkish respondents from former OHAL provinces have more positive attitudes towards various groups of Syrian refugees, regardless of the refugees’ ethnic or sectarian identity (Getmansky, Sinmazdemir, Zeitzoff, 2018).

Table 2: Determinants of Support for Public Assistance to Syrian Refugees
Health Education Housing Employment
OHAL province 0.298***








People help strangers 0.041








Kurdish 0.116***






AKP supporter 0.073***






CHP supporter -0.128***








Daily exposure to refugees -0.080**








Urban area -0.051*








Educated -0.033








Wealth 0.019








Age -0.024**








Female -0.013








Religious 0.044**








Ramadan 0.068








Observations 1023 1025 1023 1022
R-squared 0.37 0.40 0.38 0.41

Dependent variable is support for public assistance to Syrian refugees in health (column 1), education (column 2), housing (column 3), employment (column 4). All regressions include province dummies. Standard errors in parentheses. Significance: * 10%, ** 5%, *** 1%.

A potential concern with this result is that our measure of political violence at the province-level (OHAL) is too coarse. We also estimate our model using the province-level PKK fatalities data in Tezcur (2015). In regressions not reported here, we show that respondents from provinces with higher number of PKK fatalities continued to support public assistance to Syrians in all four areas.

Looking at the impact of the rest of the variables, respondents who have more frequent interaction with refugees reported lower support for assistance and, in the case of health, significantly so. But because refugee exposure is a self-reported measure, we are uncertain whether it is the refugee exposure that leads to lower support or those who oppose support are more likely to notice refugees and report interacting with them.

Kurdish respondents were also more supportive of public assistance to Syrians in health, housing and employment. However, Kurdish respondents were not more supportive of public assistance in education. One plausible explanation is that the Kurds may have interpreted this question as permission by the Turkish state for Syrians’ education in their native language of Arabic in temporary education centers. Kurds in Turkey have long demanded education in Kurdish language (Yeğen, 2015, 13). Therefore, they may have interpreted the possibility of Syrians receiving education in Arabic as a double standard of the government, which does not permit education in Kurdish in Turkish public schools (Kirişçi, 2014, 25).

Across all four different sectors of assistance, AKP supporters were more supportive of public assistance to Syrians, while CHP supporters were less supportive. Leading members of the AKP government have repeatedly referred to their “open-door” policy towards the Syrian refugees as a humanitarian duty to assist those fleeing from the violence of a cruel regime.[9] CHP leadership has been more critical of the Syrians and called for sending Syrian refugees back as part of election promises.[10] Our results indicate that the views of the voters are strongly aligned with the announced policy positions of their party.


This memo examines Turkish citizens’ attitudes towards social policies for Syrian refugees with a survey conducted in 2014. Turkey is estimated to have spent 30 billion dollars on assisting Syrians, with health and education the two largest expenditures.[11] Social policies in these areas include free healthcare services in public hospitals and free education in public schools. The results of our survey show that respondents from conflict-affected areas of Turkey exhibit greater support for assisting Syrians in health and education. One plausible explanation of this finding is that locals exposed to political violence in the past are more likely to sympathize with Syrian refugees who are also fleeing from political violence. In addition, AKP supporters and minority Kurds are more likely to support assisting Syrians. Our findings imply that segments of host societies can display altruism towards refugees even when such altruism implies a significant drain on public resources and potentially has significant economic costs for the host society.

[1]We appreciate the feedback we received from the participants of the Project on Middle East Political Science Workshop (POMEPS) on April 20, 2018 in Cambridge, MA.

[2] The numbers of registered Syrians in Turkey as of June 7th 2018 are available at http://www.goc.gov.tr/icerik6/gecici-koruma_363_378_4713_icerik

[3] The number of Syrians in these camps and their locations are available at https://www.afad.gov.tr/upload/Node/2374/files/18_06_2018_Suriye_GBM_Bilgi_Notu.pdf

[4] http://multeciler.org.tr/turkiyede-calisma-izni-verilen-suriyeli-sayisi/

[5] A more detailed description of the survey, including the design, sampling, and the instrument is available in Getmansky, Sinmazdemir, Zeitzoff (2018).

[6] Turkey is a unitary state divided into 81 provinces. Each province is composed of districts.

[7] The survey included an experiment designed to test the impact of messages about refugees on respondents’ attitudes towards the refugees as well as Turkish domestic policy and foreign policy. None of these messages had a significant impact on support for public assistance to Syrians.

[8] The exact wording of the question is “Do you agree or disagree that the Syrians use services in health, education, housing, and employment paid by the Turkish state?”

[9] See for instance, http://www.sbs.com.au/news/article/2014/04/22/almost-1-million-syrian-refugees-turkey for statements by the then Prime Minister Tayyip Erdogan.

[10] See https://www.dailysabah.com/politics/2015/04/23/chps-latest-election-promise-of-sending-back-syrian-refugees-in-turkey-comes-under-criticism

[11] https://www.gazeteduvar.com.tr/ekonomi/2017/12/05/basbakan-yardimcisi-akdagdan-maliyet-analiziyle-suriyeli-yaniti/


Aras, Bülent and Salih Yasun. 2016. The Educational Opportunities and Challenges of Syrian Refugee Students in Turkey: Temporary Education Centers and Beyond. Istanbul Policy Center-Mercator Policy Brief. Available at http://ipc.sabanciuniv.edu/wp-content/uploads/2016/07/Bulent-Aras-and-Salih-Yasun-1.pdf

Bauer, Michal, Christopher Blattman, Julie Chytilova, Joseph Henrich, Edward Miguel and Tamar Mitts. 2016. “Can War Foster Cooperation?” Journal of Economic Perspectives 30(3): 249–74.

Erdoğan, Murat. 2017. Syrians’ Barometer-2017. Available at https://mmuraterdogan.files.wordpress.com/2016/06/syrians-barometer-executive-summary.pdf.

Getmansky, Anna, Tolga Sınmazdemir and Thomas Zeitzoff. 2018. “Refugees, Xenophobia, and Domestic Conflict: Evidence from a Survey Experiment in Turkey.” Journal of Peace Research 55 (4): 491-507.

Kirişçi, Kemal. 2014. Syrian Refugees and Turkey’s Challenges: Going Beyond Hospitality. Washington, DC: Brookings Institution. Available at https://www.brookings.edu/wp- content/uploads/2016/06/Syrian-Refugees-and-Turkeys-Challenges-May- 14-2014.pdf.

Tezcur, Güneş Murat. 2015. “Electoral Behavior in Civil Wars: The Kurdish Conflict in Turkey.” Civil Wars 17(1): 70–88.

Yeğen, Mesut. 2015. “The Kurdish Peace Process in Turkey: Genesis, Evolution and Prospects.” Roma: Istituto Affari Internazionali. Working paper 11; Available at http://www.iai.it/ sites/default/files/gte_wp_11.pdf.


Table 3: Our Sample
Province District Camp in province Camp in district Respondents Kurds OHAL or adjacent zone
Adana Saricam 1 1 40 4 0
Adana Yuregir 1 0 59 13 0
Adiyaman Celikhan 1 0 30 27 1
Adiyaman Kahta 1 0 69 69 1
Adiyaman Merkez 1 1 40 37 1
Diyarbakir Baglar 0 0 60 56 1
Diyarbakir Ergani 0 0 30 30 1
Elazig Merkez 0 0 45 1 1
Gaziantep Oguzeli 1 0 30 4 0
Gaziantep Sahinbey 1 0 59 12 0
Hatay Altinozu 1 1 70 2 0
Hatay Antakya 1 1 50 1 0
Kahramanmaras Dulkadiroglu 1 1 42 2 0
Kahramanmaras Pazarcik 1 0 27 0 0
Kahramanmaras Turkoglu 1 0 29 1 0
Kayseri Hacilar 0 0 18 1 0
Kayseri Kocasinan 0 0 40 2 0
Kayseri Sarioglan 0 0 30 0 0
Kilis Elbeyli 1 1 30 3 0
Kilis Merkez 1 1 30 2 0
Mardin Kiziltepe 1 0 24 24 1
Mardin Midyat 1 1 31 17 1
Mardin Nusaybin 1 1 31 28 1
Mardin Yesilli 1 0 48 22 1
Mersin Mezitli 0 0 40 2 0
Mersin Tarsus 0 0 20 12 0
Mus Malazgirt 0 0 40 40 1
Osmaniye Merkez 1 1 30 1 0
Sanliurfa Karakopru 1 0 30 27 0
Sanliurfa Viransehir 1 1 40 36 0
Siirt Sirvan 0 0 30 30 1
Sivas Yildizeli 0 0 30 0 0
Van Catak 0 0 20 20 1