系统性风险、网络传染与金融机构系统重要性评估
|
摘要
以往国际监管组织的系统重要性金融机构测评指标体系偏重于银行群落与机构规模特征,影响了中国金融监管及理论研究的方向。笔者运用复杂网络模型、格兰杰因果检验及产品空间等方法构建中国金融风险传染网络,可视化系统性风险爆发后的风险传染路径,并通过节点出度、传染轮次、K-核分解值、LeaderRank值等指标综合评估各金融机构风险网络传染的速度、范围、深度及风险累积程度。结果表明,中国金融风险传染网络呈现出多层次、多通道的复杂关联,部分非银行金融机构在风险积聚与跨群落传导过程中发挥关键作用,规模与风险传染能力并非完全线性相关,机构关联性及负面信息传播速度等成为引发系统性风险的重要因素。打破既有群落式监管模式,推进综合监管模式创新,对当前金融深化过程中系统性风险预警与管理具有重要意义。
Systemically Important Financial Institutions( SIFIs) evaluation indexes used by international organization of supervision usually emphasis on community's structure and scale character,which influences Chinese financial supervision and direction of theoretical research. This paper constructs a model of financial risk contagion network structure by using the complex network model, Granger causality test and product space method,visualizes the risk contagion path after the outbreak of systemic risk,through which we investigate risk contagion in several aspects,include the speed,range,depth and risk accumulation degree,using the index such as node degree,contagion rounds,k-core decomposition value,leader rank value. The result shows: China's financial structure shows a multi level complex relationship other than the community structure characteristics,several non-bank financial institutions play a key role on risk accumulation and cross community risk transmission process,scale and risk contagion ability don't show completely linear correlation,institution relationship and negative information transmission speed become important factors of the systemic risk. Thus,in order to reduce risk infection in financial system,it is an inevitable choice to break the existing community supervision mode,and establish a network supervision mode.
Systemically Important Financial Institutions( SIFIs) evaluation indexes used by international organization of supervision usually emphasis on community's structure and scale character,which influences Chinese financial supervision and direction of theoretical research. This paper constructs a model of financial risk contagion network structure by using the complex network model, Granger causality test and product space method,visualizes the risk contagion path after the outbreak of systemic risk,through which we investigate risk contagion in several aspects,include the speed,range,depth and risk accumulation degree,using the index such as node degree,contagion rounds,k-core decomposition value,leader rank value. The result shows: China's financial structure shows a multi level complex relationship other than the community structure characteristics,several non-bank financial institutions play a key role on risk accumulation and cross community risk transmission process,scale and risk contagion ability don't show completely linear correlation,institution relationship and negative information transmission speed become important factors of the systemic risk. Thus,in order to reduce risk infection in financial system,it is an inevitable choice to break the existing community supervision mode,and establish a network supervision mode.
引文
[1]IMF,FSB,BIS.Guidance to Assess the Systemic Importance of Financial Institutions,Markets and Instruments:Initial Considerations[R/OL].(2009-10-28)[2015-10-15].http://www.fsb.org/wp-content/uplads/r-091107c.pdf?page_mored=1,2009.
[2]Drehmann M,Tarashev N.Measuring the Systemic Importance of Interconnected Banks[J].Journal of Financial Intermediation,2011,22(4):586-607.
[3]Franklin A,Babus A,Carletti E.Financial Connections and Systemic Risk[R].NBER Working Paper,2010:No.16177.
[4]严兵,张禹,王振磊.中国系统重要性银行评估——基于14家上市银行数据的研究[J].国际金融研究,2013(2):47-57.
[5]马君潞,范小云,曹元涛.中国银行间市场双边传染的风险估测及其系统性特征分析[J].经济研究,2007(1):68-78.
[6]Kaminsky G L,Végh C A.The Unholy Trinity of Financial Contagion[R].NBER Working Papers,2003(4):51-74.
[7]巴曙松,左伟,朱元倩.金融网络及传染对金融稳定的影响[J].财经问题研究,2013(2):3-11.
[8]Adrian T,Brunnermeier M.Co VaR[R].NBER Working Paper,No.17454,2011.
[9]Gravelle T,Li F.Measuring Systemic Importance of Financial Institutions:An Extreme Value Theory Approach[J].Journal of Banking and Finance,2013,37(7):2196-2209.
[10]温博慧,李向前,袁铭.中国非银行金融机构系统重要性再评估——基于风险倍率扩增综合指标[J].国际金融研究,2014(10):53-63.
[11]Zhou C.Are Banks Too Big to Fail?Measuring Systemic Importance of Financial Institutions[J].International Journal of Central Banking,2010,12(6):205-250.
[12]Kodres L E,Pritsker M.A Rational Expectations Model of Financial Contagion[J].Journal of Finance,2002,57(2):769-799.
[13]Kitsak M,Gallos L K,Havlin S,et al.Identification of Influential Spreaders in Complex Networks[J].Nature Physics,2010,6(11):888-893.
[14]Gai P,Kapadia S.Contagion in Financial Networks[J].Proceedings of the Royal Society A Mathematical Physical&Engineering Sciences,2010,466(2120):2401-2423.
[15]高波,任若恩.基于Granger因果网络模型的金融机构系统重要性评估[J].管理评论,2013(6):3-10.
[16]Furfine C H.Interbank Exposures:Quantifying the Risk of Contagion[J].Journal of Money Credit&Banking,2003,35(1):111-128.
[17]贾彦东.金融机构的系统重要性分析——金融网络中的系统风险衡量与成本分担[J].金融研究,2011(10):17-33.
[18]Plerou V,Parameswaran G,Bernd R.Random Matrix Approach to Cross Correlations in Financial Data[J].Physical Review E,2002,65(6):104-130.
[19]Hidalgo C A,et al.The Product Space Conditions the Development of Nations[J].Science,2007(7):482-487.
[20]BCBS.Global Systemically Important Banks:Assessment Methodology and the Additional Loss Absorbency Requirement[EB/OL].(2013-07-03)[2015-10-15].http://www.bis.org/publ/bcbs255.htm,2013.
[21]Garas A,Schweitzer F,Havlin S.A K-Shell Decomposition Method for Weighted Networks[J].New Journal of Physics,2012(8):83030-83043.
(1)ρij=i,rj)E==,其中:rσi(t)-ln Pi(t-1)表示金融机构i的股票对数收益率,Pi(t)iσjσiσj槡Var(ri)Var(rj)i(t)=ln P为股票i在第t个交易日的收盘复权价格(i=1,2,…,N,N为金融机构样本总数;t=1,2,…,T,T为统计样本股票有效交易总天数);E(·)表示统计期望平均;cov(·)是协方差;Var(·)是方差;σ是标准差。ρij∈[-1,1],若ρij>0,表示两只股票正相关;ρij<0,表示两只股票负相关;ρij=0,表示两只股票不相关;ρij=1或-1,表示两只股票完全正相关或者完全负相关。
(2)本文使用PG+方法去噪,详细过程参见Plerov等(2002)18。
(1)节点度值d'i=d-(vi)×b-槡(vi),其中b-(vi)=∑wij,是有向含权网络W中节点vi的出强度,其含义为所有以节点i为起始点的边权j=1重之和。d'i实际上综合考虑了节点vi的出度与出强度值。
(2)LeaderRank算法原本是衡量网页重要程度的方法,其核心思想是网络中一个页面的重要程度取决于指向它的页面的数量和质量。
(1)按证监会行业分类统计,截至2015年03月31日,沪深两市A股共有上市金融机构47家。其中,中国农业银行于2010年7月上市,其后两年内光大银行、兴业证券、山西证券、方正证券、国海证券、东吴证券、新华保险也相继上市。为使金融网络能涵盖更多的金融机构,故以上述最晚一家新华保险2011年12月16日上市日为样本数据起始日期。由于西部证券、中航资本、国信证券、申万宏源、东兴证券和东方证券6家机构上市较晚,与样本数据时期不一致,且都是小证券等机构,按金融稳定理事会和巴塞尔委员会发布的系统重要性金融机构评估指引,规模是其中很重要的一项,故本文综合考虑将上述6只股票剔除。
(2)限于篇幅,本文并未列示去噪前后相关系数矩阵信息,感兴趣的读者可以向作者索取。
[2]Drehmann M,Tarashev N.Measuring the Systemic Importance of Interconnected Banks[J].Journal of Financial Intermediation,2011,22(4):586-607.
[3]Franklin A,Babus A,Carletti E.Financial Connections and Systemic Risk[R].NBER Working Paper,2010:No.16177.
[4]严兵,张禹,王振磊.中国系统重要性银行评估——基于14家上市银行数据的研究[J].国际金融研究,2013(2):47-57.
[5]马君潞,范小云,曹元涛.中国银行间市场双边传染的风险估测及其系统性特征分析[J].经济研究,2007(1):68-78.
[6]Kaminsky G L,Végh C A.The Unholy Trinity of Financial Contagion[R].NBER Working Papers,2003(4):51-74.
[7]巴曙松,左伟,朱元倩.金融网络及传染对金融稳定的影响[J].财经问题研究,2013(2):3-11.
[8]Adrian T,Brunnermeier M.Co VaR[R].NBER Working Paper,No.17454,2011.
[9]Gravelle T,Li F.Measuring Systemic Importance of Financial Institutions:An Extreme Value Theory Approach[J].Journal of Banking and Finance,2013,37(7):2196-2209.
[10]温博慧,李向前,袁铭.中国非银行金融机构系统重要性再评估——基于风险倍率扩增综合指标[J].国际金融研究,2014(10):53-63.
[11]Zhou C.Are Banks Too Big to Fail?Measuring Systemic Importance of Financial Institutions[J].International Journal of Central Banking,2010,12(6):205-250.
[12]Kodres L E,Pritsker M.A Rational Expectations Model of Financial Contagion[J].Journal of Finance,2002,57(2):769-799.
[13]Kitsak M,Gallos L K,Havlin S,et al.Identification of Influential Spreaders in Complex Networks[J].Nature Physics,2010,6(11):888-893.
[14]Gai P,Kapadia S.Contagion in Financial Networks[J].Proceedings of the Royal Society A Mathematical Physical&Engineering Sciences,2010,466(2120):2401-2423.
[15]高波,任若恩.基于Granger因果网络模型的金融机构系统重要性评估[J].管理评论,2013(6):3-10.
[16]Furfine C H.Interbank Exposures:Quantifying the Risk of Contagion[J].Journal of Money Credit&Banking,2003,35(1):111-128.
[17]贾彦东.金融机构的系统重要性分析——金融网络中的系统风险衡量与成本分担[J].金融研究,2011(10):17-33.
[18]Plerou V,Parameswaran G,Bernd R.Random Matrix Approach to Cross Correlations in Financial Data[J].Physical Review E,2002,65(6):104-130.
[19]Hidalgo C A,et al.The Product Space Conditions the Development of Nations[J].Science,2007(7):482-487.
[20]BCBS.Global Systemically Important Banks:Assessment Methodology and the Additional Loss Absorbency Requirement[EB/OL].(2013-07-03)[2015-10-15].http://www.bis.org/publ/bcbs255.htm,2013.
[21]Garas A,Schweitzer F,Havlin S.A K-Shell Decomposition Method for Weighted Networks[J].New Journal of Physics,2012(8):83030-83043.
(1)ρij=i,rj)E==,其中:rσi(t)-ln Pi(t-1)表示金融机构i的股票对数收益率,Pi(t)iσjσiσj槡Var(ri)Var(rj)i(t)=ln P为股票i在第t个交易日的收盘复权价格(i=1,2,…,N,N为金融机构样本总数;t=1,2,…,T,T为统计样本股票有效交易总天数);E(·)表示统计期望平均;cov(·)是协方差;Var(·)是方差;σ是标准差。ρij∈[-1,1],若ρij>0,表示两只股票正相关;ρij<0,表示两只股票负相关;ρij=0,表示两只股票不相关;ρij=1或-1,表示两只股票完全正相关或者完全负相关。
(2)本文使用PG+方法去噪,详细过程参见Plerov等(2002)18。
(1)节点度值d'i=d-(vi)×b-槡(vi),其中b-(vi)=∑wij,是有向含权网络W中节点vi的出强度,其含义为所有以节点i为起始点的边权j=1重之和。d'i实际上综合考虑了节点vi的出度与出强度值。
(2)LeaderRank算法原本是衡量网页重要程度的方法,其核心思想是网络中一个页面的重要程度取决于指向它的页面的数量和质量。
(1)按证监会行业分类统计,截至2015年03月31日,沪深两市A股共有上市金融机构47家。其中,中国农业银行于2010年7月上市,其后两年内光大银行、兴业证券、山西证券、方正证券、国海证券、东吴证券、新华保险也相继上市。为使金融网络能涵盖更多的金融机构,故以上述最晚一家新华保险2011年12月16日上市日为样本数据起始日期。由于西部证券、中航资本、国信证券、申万宏源、东兴证券和东方证券6家机构上市较晚,与样本数据时期不一致,且都是小证券等机构,按金融稳定理事会和巴塞尔委员会发布的系统重要性金融机构评估指引,规模是其中很重要的一项,故本文综合考虑将上述6只股票剔除。
(2)限于篇幅,本文并未列示去噪前后相关系数矩阵信息,感兴趣的读者可以向作者索取。