Analysis the Factors Affecting Gold Price Fluctuations Using a Markov Switching Regression Models and Neural Networks

Gold is a strategic commodity and its price is influenced by many factors. Neural network method has a special ability to forecast ad good fitness, and markov switching regression has a special ability to distinguish the shocks and regime that switch and the exact date of fluctuations. In this study the proper model for gold price fluctuation is modeling after identifying the factor that affect on gold price in the period of 1980-2011. For modeling we use two non linear method, neural network and Markov Switching regression. The goal of this paper is not comparing the result of two method but also is better modeling and better forecasting with each model separately. Results The neural network in this study has two layers and switching regression has two regimes. The results indicate that the neural network methods are well able to predict fluctuations in the gold price. Switching regression can identify the shocks during switching yeas. And it is diagnosed that the period of being in low volatility state in the gold market is more of a high volatility state. The results show that in neural network model, among factors affecting the gold price, exchange rate and the world price of gold have the greatest impact and in Markov models, CPI has the highest importance and influence on the gold price.