Culture, Socio-Economic Development, and Refugee Immigration: A Spatial Analysis of the 2017 Referendum in Turkey

Sunum, ss.1-23, 2019

The 2017 referendum on controversial constitutional amendments witnessed a erce

competition in Turkey. Despite the joint campaign of AK Party (current ruling party)

and MHP (nationalist party), the electoral outcome yielded only a slight edge for ac-

cepting the amendments (Yes 51%, No 49%). Why was there such a narrow margin

of victory? What explains the defection among MHP voters at the aggregate level?

Our paper examines these questions through a unique dataset along with spatially-

autoregressive and multilevel modeling techniques. We collect the sub-provincial and

provincial level electoral results since 2002, and match them with the 2004 socioeco-

nomic development data from the Ministry of Development and the 2017 development

data from the Ministry of Health. In addition, we add provincial level mosque infor-

mation, and sub-provincial level ocial Syrian refugee numbers to the dataset. The

advanced geospatial and multilevel models show strong empirical support for our hy-

potheses. Cultural indicators are as likely as socioeconomic features to explain the 2017

referendum results in Turkey after controlling for political factors. The number of Syr-

ian refugees in provinces diminishes the level of support for \yes" whereas the number

of mosques boosts its support. The refugee migration also explains the MHP voters

who defected in the referendum: the higher the number of refugees in a sub-province,

the more likely that the voters in that sub-province vote against the constitutional

amendments. All models show that voting in Turkey has considerable levels of spatial

dependency - the neighborhood matters.