价格惯性、波动性与学习型预期——以农产品和能源价格为例的研究

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

针对现有测度惯性模型的不足,本文将常系数惯性模型推广到含体制变化的Markov模型,并将随机波动、学习型预期等引入到惯性模型中。论文首先利用最小二乘法、分位数回归、贝叶斯方法、未知断点检验法、Markov模型研究了农产品价格和能源价格的惯性特征,然后利用GARCH和SV模型测度了农产品价格和能源价格的波动性特征,继而研究了农产品价格和能源价格的学习型预期,最后研究了包含惯性、波动性和学习型预期的动态模型。实证研究表明:能源价格的惯性强于农产品价格惯性;农产品价格惯性模型比能源价格惯性模型稳定;相对于GARCH模型,随机波动(SV)模型能够更好地测度农产品价格和能源价格的波动性;农产品价格水平与波动性之间互相影响,而能源价格水平与波动性没有影响;能源价格预期模型的最优学习速率大于农产品价格预期模型的最优学习速率;农产品价格和能源价格的学习型预期都不是其惯性来源,对其价格形成的影响有限。

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.