商业银行压力测试宏观情景构建及应用——基于FAVAR模型
|
摘要
本文遵循国际最新监管要求并结合商业银行主要风险来源和自身经营模式,利用FAVAR模型构建包含68个风险因子的未来九个季度宏观情景。模型预测结果显示,当关键风险因子不出现结构性变动时,模型表现出极强的预测能力;当关键风险因子出现结构性变动时,模型预测能力会出现减弱迹象。此外,模型生成的宏观情景除可通过进一步传导广泛运用于风险损失、拨备前净利润预测外,还有助于提升银行对经济资本及监管资本的主动管理能力,有效推进银行全面风险管理体系的构建。
The authors of this paper follow the latest international regulatory requirements, consider the main risk sources and business models of commercial banks, and use the FAVAR model to construct macro-scenarios with 68 risk factors in the following nine quarters. The results of the model prediction show that, if the key risk factors do not structurally change,the model has a strong predictive ability; if the key risk factors structurally change, the predictive ability of the model will be weakened. In addition, the macro-scenario generated by the model not only can be widely used to predict the risk loss and the pre-provision net profits, but also helps improve the bank's ability in active management of economic capitals and regulatory capitals, and effectively promote the construction of the comprehensive risk management system of banks.
The authors of this paper follow the latest international regulatory requirements, consider the main risk sources and business models of commercial banks, and use the FAVAR model to construct macro-scenarios with 68 risk factors in the following nine quarters. The results of the model prediction show that, if the key risk factors do not structurally change,the model has a strong predictive ability; if the key risk factors structurally change, the predictive ability of the model will be weakened. In addition, the macro-scenario generated by the model not only can be widely used to predict the risk loss and the pre-provision net profits, but also helps improve the bank's ability in active management of economic capitals and regulatory capitals, and effectively promote the construction of the comprehensive risk management system of banks.
引文
程功、文晴、李志勇,2017.基于时变参数模型的银行信用风险与宏观经济形势的动态关系研究[J].金融监管研究,(3):51-72.
贺力平、谈俊,2016.债务率与经济周期的关系[J].金融论坛,(11):3-11.
谭晓红、樊纲治,2011.我国商业银行宏观压力测试研究——基于四类银行的SUR模型[J].投资研究,(12):3-16.
汤婷婷、方兆本,2011.商业银行信用风险与宏观经济——基于压力测试的研究[J].当代经济科学,(4):66-71.
张棋、王玥、李鑫,2012.信贷资产质量前瞻性预测与压力测试——基于ARMA模型和VAR模型的研究[J].金融论坛,(5):19-25.
Bai,J.,and Ng,S.,2002.Determining the Number of Factors in Approximate Factor Models[J].Econometrica,70(1):191-221.
Bernanke,B.S.,Boivin,J.,and Eliasz,P.,2005.Measuring the Effects of Monetary Policy:A Factor-augmented Vector Autoregressive(FAVAR)Approach[J].The Quarterly Journal of Economics,120(1):387-422.
Cao,J.,and Chollete,L.,2014.Capital Adequacy and Liquidity in Banking Dynamics[R].UIS Working Papers in Economics and Finance.
Covas,F.B.,Rump,B.,and Zakraj觢ek,E.,2014.Stress-testing US Bank Holding Companies:A Dynamic Panel Quantile Regression Approach[J].International Journal of Forecasting,30(3):691-713.
Duan,J.C.,and Zhu,Y.Q.,2018.Dynamic Macro Scenario Analysis via Bridge Sampling[R].Available at SSRN:https://ssrn.com/abstract=3055126.
Flavin,T.,and Lagoa-Varela D.,2016.Are Banking Shocks Contagious?Evidence from the Eurozone[R].2016.Working Paper N268-16.
Gupta,R.,Jurgilas,M.,and Kabundi,A.,2010.The Effect of Monetary Policy on Real House Price Growth in South Africa:AFactor-augmented Vector Autoregression(FAVAR)Approach[J].Economic Modelling,27(1):315-323.
Hoggarth,G.,Sorensen,S.,and Zicchino,L.,2005.Stress Tests of UK Banks Using a VAR Approach[R].Bank of England Working Paper Series No.282.
Hosszú,Z.,2018.The Impact of Credit Supply Shocks and a New Financial Conditions Index Based on a FAVAR Approach[J].Economic Systems,42(1):32-44.
Kolasa,M.,2009.Structural Heterogeneity or Asymmetric Shocks?Poland and the Euro Area through the Lens of a Two-country DSGE Model[J].Economic Modelling,26(6):1245-1269.
Louzis,D.P.,Vouldis,A.T.,and Metaxas,V.L.,2012.Macroeconomic and Bank-specific Determinants of Non-performing Loans in Greece:A Comparative Study of Mortgage,Business and Consumer Loan Portfolios[J].Journal of Banking&Finance,36(4):1012-1027.
Sorge,M.,2004.Stress-testing Financial Systems:An Overview of Current Methodologies[R].BIS Working Paper No.165.
Stock,J.H.,and Watson,M.W.,2002.Macroeconomic Forecasting Using Diffusion Indexes[J].Journal of Business&Economic Statistics,20(2):147-162.
(1)美联储2009年所进行的压力测试为监管资本评估项目(Supervisory Capital Assessment Program,SCAP)。随着《多德一弗兰克法案》的颁布,美联储将该压力测试项目自2010年起更名为全面资本评估与审查(Comprehensive Capital Analysis and Review,CCAR)。
(1)以一阶滞后的n变量VAR模型为例,待估参数数目为n(n+1),这将需要进一步增加观测值数量才能保证参数估计的精度,Gupta等(2010)认为VAR模型最多能容纳12个变量,超过该数目后模型将失去实际预测能力。
(1)“快速变化”的经济指标表示压力环境下对宏观经济环境变动敏感的指标,如利率、美元兑在岸人民币1年期隐含波动率等指标;“慢速变化”的经济指标表示压力环境下对宏观经济环境变动迟钝的指标,如产出、消费及价格等指标。
(2)全面性指所选取的因子应涵盖财务预测、风险预测,并考虑到国内、国际多方面因素的影响,以便全方位反映银行实际承压情况;可获得性指由于因子最终是作为后续压力测试的自变量,需考虑获取定量因子的可行性;可解释性指所选取的因子要具有现实和合理。
(1)外生变量选择范围综合参考了CCAR和香港金融管理局压力测试要求。外生变量选择上,香港最优贷款利率等11个变量为香港金融管理局压力测试中要求的监管变量,美国GDP增速为美联储CCAR中要求的监管变量。
(1)按照2017年CCAR压力测试要求,本次压力测试窗口期为九个季度。同样,下文中综合2017年CCAR和中国人民银行FASP压力测试要求,将表1中的中国名义GDP增速、固定资产投资完成额(累计同比)、出口金额、CPI增幅、美元兑在岸人民币即期汇率、M2同比增速、中国香港GDP增速共7个变量作为关键风险因子。
(2)本文所提取的7个主成分对中国主要宏观经济变量的解释效果显著。从主成分波动轨迹看,中国经济结构调整较大的2002-2003年、2009-2010年间,7个主成分的波动幅度也较为明显;此外,从累积方差贡献度来看,7个主成分对于中国名义GDP增速、固定资产投资完成额(累计同比)、出口金额、CPI增幅、美元兑在岸人民币即期汇率、M2同比增速这一类中国内地主要宏观经济指标的累计方差贡献度分别高达83.12%、82.17%、76.38%、80.19%、78.81%、84.56%,有效刻画了关键风险因子的运行特征。
(3)宏观情景生成68个变量在不同情景下的预测路径时,突出7个关键风险因子主要是因为这7个变量在整个压力测试后续环节中有重要应用,7个变量预测路径生成后银行会对预测值进行专家判断,最后以修正后的预测路径对本组风险指标体系进行小幅校准,最终使得68个变量生成的情景更切合实际。
(4)对于7组风险指标体系分组规则参见表1。对于资产价格变量组,由于顺周期条件下固定资产投资累计增速与该组的五个变量平均相关系数0.63,因此以固定资产投资累计增速为参照系,根据联合距离函数来判定资产价格变量组的基准情景、不利情景、严重不利情景的生成。
贺力平、谈俊,2016.债务率与经济周期的关系[J].金融论坛,(11):3-11.
谭晓红、樊纲治,2011.我国商业银行宏观压力测试研究——基于四类银行的SUR模型[J].投资研究,(12):3-16.
汤婷婷、方兆本,2011.商业银行信用风险与宏观经济——基于压力测试的研究[J].当代经济科学,(4):66-71.
张棋、王玥、李鑫,2012.信贷资产质量前瞻性预测与压力测试——基于ARMA模型和VAR模型的研究[J].金融论坛,(5):19-25.
Bai,J.,and Ng,S.,2002.Determining the Number of Factors in Approximate Factor Models[J].Econometrica,70(1):191-221.
Bernanke,B.S.,Boivin,J.,and Eliasz,P.,2005.Measuring the Effects of Monetary Policy:A Factor-augmented Vector Autoregressive(FAVAR)Approach[J].The Quarterly Journal of Economics,120(1):387-422.
Cao,J.,and Chollete,L.,2014.Capital Adequacy and Liquidity in Banking Dynamics[R].UIS Working Papers in Economics and Finance.
Covas,F.B.,Rump,B.,and Zakraj觢ek,E.,2014.Stress-testing US Bank Holding Companies:A Dynamic Panel Quantile Regression Approach[J].International Journal of Forecasting,30(3):691-713.
Duan,J.C.,and Zhu,Y.Q.,2018.Dynamic Macro Scenario Analysis via Bridge Sampling[R].Available at SSRN:https://ssrn.com/abstract=3055126.
Flavin,T.,and Lagoa-Varela D.,2016.Are Banking Shocks Contagious?Evidence from the Eurozone[R].2016.Working Paper N268-16.
Gupta,R.,Jurgilas,M.,and Kabundi,A.,2010.The Effect of Monetary Policy on Real House Price Growth in South Africa:AFactor-augmented Vector Autoregression(FAVAR)Approach[J].Economic Modelling,27(1):315-323.
Hoggarth,G.,Sorensen,S.,and Zicchino,L.,2005.Stress Tests of UK Banks Using a VAR Approach[R].Bank of England Working Paper Series No.282.
Hosszú,Z.,2018.The Impact of Credit Supply Shocks and a New Financial Conditions Index Based on a FAVAR Approach[J].Economic Systems,42(1):32-44.
Kolasa,M.,2009.Structural Heterogeneity or Asymmetric Shocks?Poland and the Euro Area through the Lens of a Two-country DSGE Model[J].Economic Modelling,26(6):1245-1269.
Louzis,D.P.,Vouldis,A.T.,and Metaxas,V.L.,2012.Macroeconomic and Bank-specific Determinants of Non-performing Loans in Greece:A Comparative Study of Mortgage,Business and Consumer Loan Portfolios[J].Journal of Banking&Finance,36(4):1012-1027.
Sorge,M.,2004.Stress-testing Financial Systems:An Overview of Current Methodologies[R].BIS Working Paper No.165.
Stock,J.H.,and Watson,M.W.,2002.Macroeconomic Forecasting Using Diffusion Indexes[J].Journal of Business&Economic Statistics,20(2):147-162.
(1)美联储2009年所进行的压力测试为监管资本评估项目(Supervisory Capital Assessment Program,SCAP)。随着《多德一弗兰克法案》的颁布,美联储将该压力测试项目自2010年起更名为全面资本评估与审查(Comprehensive Capital Analysis and Review,CCAR)。
(1)以一阶滞后的n变量VAR模型为例,待估参数数目为n(n+1),这将需要进一步增加观测值数量才能保证参数估计的精度,Gupta等(2010)认为VAR模型最多能容纳12个变量,超过该数目后模型将失去实际预测能力。
(1)“快速变化”的经济指标表示压力环境下对宏观经济环境变动敏感的指标,如利率、美元兑在岸人民币1年期隐含波动率等指标;“慢速变化”的经济指标表示压力环境下对宏观经济环境变动迟钝的指标,如产出、消费及价格等指标。
(2)全面性指所选取的因子应涵盖财务预测、风险预测,并考虑到国内、国际多方面因素的影响,以便全方位反映银行实际承压情况;可获得性指由于因子最终是作为后续压力测试的自变量,需考虑获取定量因子的可行性;可解释性指所选取的因子要具有现实和合理。
(1)外生变量选择范围综合参考了CCAR和香港金融管理局压力测试要求。外生变量选择上,香港最优贷款利率等11个变量为香港金融管理局压力测试中要求的监管变量,美国GDP增速为美联储CCAR中要求的监管变量。
(1)按照2017年CCAR压力测试要求,本次压力测试窗口期为九个季度。同样,下文中综合2017年CCAR和中国人民银行FASP压力测试要求,将表1中的中国名义GDP增速、固定资产投资完成额(累计同比)、出口金额、CPI增幅、美元兑在岸人民币即期汇率、M2同比增速、中国香港GDP增速共7个变量作为关键风险因子。
(2)本文所提取的7个主成分对中国主要宏观经济变量的解释效果显著。从主成分波动轨迹看,中国经济结构调整较大的2002-2003年、2009-2010年间,7个主成分的波动幅度也较为明显;此外,从累积方差贡献度来看,7个主成分对于中国名义GDP增速、固定资产投资完成额(累计同比)、出口金额、CPI增幅、美元兑在岸人民币即期汇率、M2同比增速这一类中国内地主要宏观经济指标的累计方差贡献度分别高达83.12%、82.17%、76.38%、80.19%、78.81%、84.56%,有效刻画了关键风险因子的运行特征。
(3)宏观情景生成68个变量在不同情景下的预测路径时,突出7个关键风险因子主要是因为这7个变量在整个压力测试后续环节中有重要应用,7个变量预测路径生成后银行会对预测值进行专家判断,最后以修正后的预测路径对本组风险指标体系进行小幅校准,最终使得68个变量生成的情景更切合实际。
(4)对于7组风险指标体系分组规则参见表1。对于资产价格变量组,由于顺周期条件下固定资产投资累计增速与该组的五个变量平均相关系数0.63,因此以固定资产投资累计增速为参照系,根据联合距离函数来判定资产价格变量组的基准情景、不利情景、严重不利情景的生成。