基于预测能力的扩展极限边界分析法  

Prediction Based Augmented Extreme Bound Analysis

作　　者：张大永[1] 邹沛江[1] 

机构地区：[1]西南财经大学经济与管理研究院

出　　处：《数量经济技术经济研究》2014年第2期99-112,共14页Journal of Quantitative & Technological Economics

摘　　要：本文在传统EBA方法的基础上,将其引入到时间序列中,构建以预测为导向的AEBA模型选择方法。AEBA在模型选择上更注重于模型的预测能力,在稳健性检验上细分为模型稳健性检验与参数稳健性检验两部分,提出了基于时间序列预测能力的检验方法。最后实证示例用AEBA方法对影响石油股票指数收益率的因素进行了研究,表明该方法选择的模型的预测能力,特别是短期预测能力要显著强于CAPM、三因子模型、ARMA以及VAR。We build a prediction based model （AEBA） to extend the traditional EBA to time series model. AEBA pays more attention to prediction based model se- lection and conduct a robustness test based on time series prediction. The empirical study in this paper uses AEBA method to study factors which affect the yield of oil stock index and use it to predict the yield of oil stock index. It shows that predictive ability of the model selected by AEBA is much better than CAPM, three-factor model, ARMA, and VAR.

关 键 词：极限边界分析 扩展极限边界分析 模型选择 稳健性检验 

分 类 号：F830[经济管理—金融学]