Modeling Price Dynamics and Risk Forecasting in Tehran Stock Exchange Market: Nonlinear and Non-gaussian Models of Stochastic Volatility

The daily observations of the total index of the Tehran Stock Exchange show that in the last few years, stock prices have been very volatile. This volatility can harm the economic environment of Iran. Modeling and predicting price volatility in this market can provide important information about uncertainty and risk to actors and thus help with managing possible unwanted changes in the field of financial investments. This, along with the rise in the value of the stock market in recent years, has caused the investigation of the issue of volatility to become increasingly popular among academicians and financial policymakers. Volatility and risk measurements are essential parameters in risk management programs and can affect a country’s economic activity and public confidence. These are also key parameters in studies that examine the relationship between the stock market, economic growth, and other financial variables. Tehran Stock Exchange markets have been volatile in recent years. Controlling the negative effects caused by stock price volatility has made it necessary to predict and model price dynamics for participants in this market. In this paper, the class of Parameter-Driven volatility models (stochastic volatility models) is used to predict the price volatility and calculate the risk of the price index in the Tehran stock market. Therefore, four stochastic volatility models were used. To make a comprehensive review, the asymmetry in the volatility (leverage effect) and the heavy tail of the stock price return distribution (with t- student distribution and Skew normal) have been included in the models. To estimate the models, the Gibbs sampling method was used, and to accurately compare the models, the test based on the posterior distribution of the models and the Bayesian factor was used. The results indicated that the canonical stochastic volatility model with Skew normal distribution (SNSV) is more effective than other stochastic volatility models in predicting the price of stock market volatility based on the Bayesian factor. Therefore, to analyze stock market risks using stochastic volatility models, there is no need to include the leverage effect in the state space of the Volatility equation. The SNSV model makes it possible to observe volatility and make predictions related to it, thereby improving market transparency and ultimately making diversification and risk management easier to implement. Also, the backtests of VaR and CVaR market risk assessment using Kupiec and DQ tests do not show evidence that the estimation is over or under the risk limit. As a result, the calculation of volatility and pricing with this model will lead to more precision risk management for professionals, especially fund managers who intend to include Tehran Stock Exchange stocks for asset allocation.