Nonlinear Dynamic Modeling of Factors Affecting the Stock Market: Baysian Quantile Threshold Regression - GARCH approach

During the last decade, studies on the factors affecting stock market returns have reached a peak with the advances of financial economics in the field of statistics and mathematics, and modeling is of great importance in this regard. Accordingly, this study seeks to present a new approach to modeling the nonlinear relationship between financial variables and stock returns. This paper employs Bayesian Markov chain Monte Carlo (MCMC) sampling methods for updating the estimates and quantile threshold regression with heteroscedasticity. To study and model this approach, we used returns of the Tehran Stock Exchange, Coin Price, Oil, and Gold Price from 2011 to 2019. The results show that these variables have different effects under low and upper quantile levels. Also, the impact of the financial variables on the stock returns is different under higher and lower threshold amount for each quantile levels. Based on the result, we can say that stock returns have a nonlinear relationship with other markets in the bullish and bearish market.