GARCH-neural网络模型对股票价格买卖差波动率的预测

股票价格买卖差是衡量金融市场流动性和有效性的重要指标,已经得到学术界的广泛研究.相比而言,作为衡量股票市场风险的重要因素的股票价格买卖价差的波动率却没有得到相同的重视.在广义自回归条件异方差(generalized autoregressive conditional heteroscedasticity,GARCH)模型的基础上,提出了GARCH-neural network (GARCH-NN)混合模型分析股票价格买卖价差波动率的动态性.以深圳证券交易所成分股价指数的高频数据为样本对所提模型进行了实证分析.运用GARCH家族模型对股票价格买卖差波动率的动态性进行分析,得出预测效果最优的GARCH模型.在最优GARCH模型的基础上结合神经网络分析方法即GARCH-NN混合模型对样本数据进行了实证分析.比较分析最优GARCH模型和GARCH-NN混合模型对股票价格买卖差波动率的预测效果,并以AIC(Akaike information criterion)和BIC(Bayesian information criterion)作为检验模型预测效果的指标.实证结果表明,提出的GARCH-NN混合模型更优.

GARCH-neural network model for forecasting the volatility of bid-ask spread of the Chinese stock market

The bid-ask spread is an important indicator to evaluate the financial market liquidity and efficiency, which has been extensively studied by recent literature. While, the volatility of bid-ask spread as one important determinant of market risk did not arouse enough attention from researchers to our knowledge. In this paper, we study the dynamics of bid-ask spread volatility of 40 constituent stocks of Shenzhen Stock Exchange Index (SZEI) in Chinese stock market, using GARCH (generalized autoregressive conditional heteroscedasticity) family models. Perfected GARCH model is identified according to AIC(Akaike information criterion) and BIC(Bayesian information criterion) criteria, then a hybrid GARCH-neural network (GARCH-NN) model based on it is proposed to forecast the volatility of SZEI bid-ask spread. Forecast performances obtained by the perfected GARCH model are compared with that by GARCH-NN based on our SZEI bid-ask spread dataset. The results show that the proposed hybrid model in this paper performs better.