The ‘Core Vote’ and ‘Swing Vote’ are two di↵erent hypotheses explaining the distribution of government resources. In this paper, we test the Core Vote (Cox and Mc-Cubbins, 1986) versus Swing Vote (Lindbeck and Weibull, 1987; Dixit and Londregan, 1996) hypotheses and develop our theory of which strategy a government would adopt as the country moves from democracy towards autocracy. We develop a formal model to show the relationship between changes in democracy and its impact on which voting block(s) receives a higher distribution of public resources. Using the data of public housing between 2002 and 2016 and four general election results (2002, 2007, 2011, 2015) in Turkey, our OLS specifications show that our estimates of co-party bias follow the Polity IV trend very closely. As such, we find that the less democratic Turkey becomes, the more co-party bias there is. We expand on this initial discovery through the use of fixed effects regressions and further use an instrumental variable estimation as our identification strategy. Our results illustrate that the change of the incumbent Justice and Development Party’s (AKP) vote share positively impacts the quantity of public housing built in a province which has a large AKP vote share.
How to Cite:
Caspi, E., Dann, C., Sun, L. and Zhu, Y., 2020. Anocracy in Ankara: Co-party Bias and the Turkish Housing Development Administration (TOKI), 2002 – 2016. Rationale, 1, pp.98–124.