HE Qizhi:Price Persistence, Volatility and Learning Expectation ——Taking Agricultural and Energy Prices for Example

Abstract：Based on the inadequacy of existing models, the constant coefficient persistence model is extended to the Markov model with regime change, and stochastic volatility and learning expectation are introduced into the persistence model.First, the persistence of agricultural price and energy price are researched using least squares method, Quantile regression, Bayesian estimate, unknown breakpoint test and Markov model respectively.Then volatility characteristics of agricultural price and energy price are measured by GARCH and SV models.Learning expectation for agricultural price and energy price is also researched.Finally dynamic models including persistence, volatility and learning expectation are researched.Empirical research shows that the persistence of energy price is higher than that of agricultural price.The persistence model of agricultural price is more stable than that of energy price.Stochastic volatility(SV) model is better than GARCH model in measuring the volatility of agricultural price and energy price.As to agricultural prices, the level term and its volatility influence each other, while energy price does not.The optimal learning rate of energy price expectation model is greater than that of agricultural price.Whether to energy price or agricultural price, learning expectation is not the source for the persistence, and it has a limited impact on the price formation.

Key words： Price    Persistence    Converting Mechanism    Stochastic Fluctuations    Learning Expectation

source:Finance & Trade Economics ,No.2,2015