中文摘要：

基于神经网络分位数CCK（以下简记为QRNN-CCK）模型的基金市场羊群效应的检验,一方面通过分位数回归揭示市场收益对CASD指标整个条件分布的影响;另一方面,通过神经网络结构模拟市场收益对CASD指标的非线性影响模式。从而很好地解决了羊群效应检验中遇到的两个难题：羊群效应的异质性和非线性关联模式。以开放式基金为研究对象,对基金市场羊群效应进行判断,实证结果表明：第一,与分位数回归-CCK模型相比,QRNN-CCK模型能获得更好的解释能力和拟合效果;第二,在各分位点处,市场收益对CSAD指标的影响都以市场收益0为分界点呈现非对称性的＂V型＂特征,表明基金市场羊群效应显著,而且在不同分位点处具有不同的变化模式,表现出异质性;第三,对基金市场的牛市和熊市进行更细致的对比发现,牛市和熊市均随着分位点的提高,市场收益对CSAD的影响越来越明显,并且牛市与熊市的表现存在差异。

英文摘要：

Based on the testing of QRNN-CCK model for open-ended fund market＇s herd behavioral effect,the influence of market benefits on the CASD index conditional distribution has been revealed by quantile regression; besides,the nonlinear influence model of market benefits simulated by neural network structure on CASD index has also been determined. In this way,the two difficult issues existing in the herd behavioral effect testing have been solved： the heterogeneity and nonlinear relation pattern of herd behavioral effect. Setting the open-ended fund as the research example and by the herd behavioral effect of fund market,the results show that,firstly,comparing to quantile regression-CCK model,QRNN-CCK can obtain strong explanatory and good fitting results; Secondly,the impact of the market return to CSAD displays the characteristics of asymmetric ＂V-type＂at each quantile site,and the cut-off point is 0,which shows that herd behavioral effect works out obviously in the fund market,while there is different changing model in different quantile; thirdly,after a more detailed comparison between the bull market and bear market of fund market,with the increase of quantile of the two markets,market effect has a more obvious influence on CSAD. At the same time,there is a difference between the performance of bull market and bear market.