The analysis of relationship between stock prices andselected economic variables in Turkish economy by nonlinear cointegration tests

In this study, threshold autoregressive models were examined as a family of nonlinear models in mean. Before using nonlinear methods in time series analysis, it is necessary to determine that the series is not linear. Therefore, the study includes the examination of linearity tests. The concept of unit root, unit root tests with structural break and nonlinear unit root tests are emphasized. Nonlinear cointegration tests, which are less seen in the econometrics literature, have been examined. In Turkish economy, the relationship between the stock price and economic variables, were investigated by using KSS (2006), Enders-Siklos (2001), Maki (2010) cointegration tests and ESTAR error correction model. According to the results, there is a cointegration with ESTAR adjustment between stock price and exchange rate, gold price, inflation rate and interest rate. Except the interest rate, there is a long-term causality between the stock price and the other variables.