Abstract
This study investigated the existence of chaotic structure in voting behavior by considering non-economic and macroeconomic factors in Turkey during the period of 03.1986-01.2020. The chaotic structure among the analyzed variables was characterized by Lyapunov exponents that explore the chaotic dynamics of the series. Following, the effects of inflation, unemployment, economic growth and terror on party votes were analyzed by Fourier regression model. Then, the causality among the macroeconomic variables, terror and party groups was analyzed by the Granger causality method. According to our results, there is unidirectional causality from terror to all four party groups. In the context of macroeconomic variables, there is the evidence of bidirectional causality between conservative parties and inflation; unidirectional causality from inflation to center-right and center-left parties. There is no causality between nationalist parties and inflation. Furthermore, center-right and center-left parties have the evidence of no causality with unemployment while there is unidirectional causality from unemployment to conservative and nationalist parties. There is unidirectional causality from economic growth to conservative parties and bidirectional causality between center-right parties and economic growth. However, the center-left and nationalist parties are not the sources of Granger causality of economic growth, and there is no inverse Granger causality relationship between these variables. Therefore, it can be concluded that between the periods 03.1986-01.2020, there was no concern for economic growth in left-wing and nationalist-based parties in Turkey.