对我国16家上市银行破产风险特征因素研究

Jarry1

于 2020-06-18 14:17:43 发布

阅读量756

点赞数 1

本文链接：https://blog.csdn.net/jarry1/article/details/106832861

版权

#季度数据
##银行特征与系统性风险
library("readr")
library("readxl")
library("dplyr")
library(stringr)
library(PerformanceAnalytics)
library(plm)
setwd("e:/R/tail risk/day/finance")
rf1<-read.csv("rf1.csv")
listf<-read_excel("listf.xlsx",sheet = 1)%>%select(-2)
rf1<-left_join(rf1,listf,by="Stkcd")
varp<-0.05
# 读取 银行上市股票代码
cs1<-read_excel("cs.xlsx",sheet = 1)%>%arrange(Stkcd)
# 读取金融指数：
findex<-read_excel("findex.xlsx",sheet = 1)%>%select(2,3,12)%>%rename(Stkcd="简称",Clsdt="日期",Dretwd="涨跌幅(%)")%>%filter(Dretwd!="--")
findex1<-data.frame(findex[,c(1,2)],Dretwd=as.numeric(findex$Dretwd))
findex1<-(data.frame(findex1,Trdmnt=as.numeric(str_replace_all(findex1$Clsdt,"-","")))%>%select(-2))[,c(1,3,2)]
cs<-filter(findex1,Stkcd=="金融(风格.中信)")%>%mutate(Cdretwdos=Dretwd/100)%>%select(-1,-3)
# 读取无风险收益率
wf<-read_excel("wf.xlsx",sheet = 1)%>%slice(-(1L:2L))%>%select(-1,-4)
wf1<-data.frame(Trdmnt=as.numeric(str_replace_all(wf$Clsdt,"-","")),Nrrdaydt=as.numeric(wf$Nrrdaydt))
# 银行收益率与金融指数收益率
rb<-semi_join(rf1,cs1,by="Stkcd")%>%inner_join(cs,by="Trdmnt")%>%left_join(wf1,by="Trdmnt")

# 开始求出银行个股2005-2018年面板数据的系统性尾部风险
d=list();d1=list();d2=list();d11=list();d12=list()
for(i in 0:9){
  for(h in 0:3){
    if(i==9&h==1) break
    r1<-filter(rb,Trdmnt>=(20050101+300*h+10000*i),Trdmnt<(20100101+300*h+10000*i))
    # 筛选出上市时期不大于5年窗口期第一个交易日的股票
    r1<-filter(r1,List<=min(r1$Trdmnt))
    k<-round(nrow(distinct(r1,Trdmnt))*varp)
    r2<-distinct(r1,Trdmnt,.keep_all = TRUE)%>%transmute(rme=Cdretwdos,rme=-rme)%>%arrange(rme)
    r3<-slice(r2,(nrow(distinct(r1,Trdmnt))-k+1):n())
    r4<-transmute(r3,lrme=log(rme));a<-sum(r4)       
    α1<-1/k*a-log(as.numeric(slice(r2,nrow(distinct(r1,Trdmnt))-k))) #得出 Hill estimator
    r5<-mutate(r1,rje=-(Dretwd),rme=-(Cdretwdos))
    b<-group_by(r5,Stkcd)%>%arrange(rme)%>%slice(n()-round(n()*varp))%>%select(Stkcd,rme)%>%rename(frme=rme) #市场VAR
    f<-group_by(r5,Stkcd)%>%arrange(rje)%>%slice(n()-round(n()*varp))%>%select(Stkcd,rje)%>%rename(frje=rje) #个股VAR
    c<-left_join(r5,f,by="Stkcd")%>%left_join(b,by="Stkcd")
    c1<-mutate(c,t=ifelse(rje>frje&rme>frme,1,0),tt=ifelse(rme>frme,1,0))
    c2<-group_by(c1,Stkcd)%>%summarise(τ=sum(t)/sum(tt)) 
    varm<-b
    varj<-f
    d[[i*4+(h+1)]]<-data.frame(left_join(c2,varj,by="Stkcd")%>%left_join(varm,by="Stkcd")%>%mutate(tβ=(τ^α1)*frje/frme)%>%select(-(2:4))%>%filter(tβ!="NA",tβ!="NaN"),Trdmnt=200501+3*h+100*i)[,c(1,3,2)]
    d1[[i*4+(h+1)]]<-data.frame(left_join(c2,varj,by="Stkcd")%>%left_join(varm,by="Stkcd")%>%mutate(crisk=τ^α1)%>%select(-(2:4))%>%filter(crisk!="NA",crisk!="NaN"),Trdmnt=200501+3*h+100*i)[,c(1,3,2)]
    d2[[i*4+(h+1)]]<-data.frame(left_join(c2,varj,by="Stkcd")%>%left_join(varm,by="Stkcd")%>%mutate(brisk=frje/frme)%>%select(-(2:4))%>%filter(brisk!="NA",brisk!="NaN"),Trdmnt=200501+3*h+100*i)[,c(1,3,2)]
    d11[[i*4+(h+1)]]<-data.frame(group_by(r1,Stkcd)%>%summarise(p=cov(Dretwd,Cdretwdos)/(sd(Dretwd)*sd(Cdretwdos)))%>%filter(p!="NA",p!="NaN"),Trdmnt=200501+3*h+100*i)[,c(1,3,2)]
    d12[[i*4+(h+1)]]<-data.frame(group_by(r1,Stkcd)%>%summarise(d=sd(Dretwd)/sd(Cdretwdos))%>%filter(d!="NA",d!="NaN"),Trdmnt=200501+3*h+100*i)[,c(1,3,2)]       
    }}
badj<-read_excel("badj.xlsx",sheet = 1);roaj<-read_excel("roaj.xlsx",sheet = 1);capj<-read_excel("capj.xlsx",sheet = 1)
netj<-read_excel("netj.xlsx",sheet = 1);costj<-read_excel("costj.xlsx",sheet = 1);otherj<-read_excel("otherj.xlsx",sheet = 1)
hdj<-read_excel("hdj.xlsx",sheet = 1);changej<-read_excel("changej.xlsx",sheet = 1);investj<-read_excel("investj.xlsx",sheet = 1)
servicej<-read_excel("servicej.xlsx",sheet = 1);netimj<-read_excel("netimj.xlsx",sheet = 1);incomej<-read_excel("incomej.xlsx",sheet = 1)
equityj<-read_excel("equityj.xlsx",sheet = 1);assetj<-read_excel("assetj.xlsx",sheet = 1);depositj<-read_excel("depositj.xlsx",sheet = 1)
loanj<-read_excel("loanj.xlsx",sheet = 1);roej<-read_excel("roej.xlsx",sheet = 1)
#
badj1<-as.matrix(badj[,3:78])%>%t();dim(badj1)<-c(76*33,1)
roaj1<-as.matrix(roaj[,3:78])%>%t();dim(roaj1)<-c(76*33,1)
capj1<-as.matrix(capj[,3:78])%>%t();dim(capj1)<-c(76*33,1)
roej1<-as.matrix(roej[,3:78])%>%t();dim(roej1)<-c(76*33,1)
netj1<-as.matrix(netj[,3:78])%>%t();dim(netj1)<-c(76*33,1)
costj1<-as.matrix(costj[,3:78])%>%t();dim(costj1)<-c(76*33,1)
otherj1<-as.matrix(otherj[,3:78])%>%t();dim(otherj1)<-c(76*33,1)
hdj1<-as.matrix(hdj[,3:78])%>%t();dim(hdj1)<-c(76*33,1)
changej1<-as.matrix(changej[,3:78])%>%t();dim(changej1)<-c(76*33,1)
investj1<-as.matrix(investj[,3:78])%>%t();dim(investj1)<-c(76*33,1)
servicej1<-as.matrix(servicej[,3:78])%>%t();dim(servicej1)<-c(76*33,1)
netimj1<-as.matrix(netimj[,3:78])%>%t();dim(netimj1)<-c(76*33,1)
incomej1<-as.matrix(incomej[,3:78])%>%t();dim(incomej1)<-c(76*33,1)
equityj1<-as.matrix(equityj[,3:78])%>%t();dim(equityj1)<-c(76*33,1)
assetj1<-as.matrix(assetj[,3:78])%>%t();dim(assetj1)<-c(76*33,1)
depositj1<-as.matrix(depositj[,3:78])%>%t();dim(depositj1)<-c(76*33,1)
loanj1<-as.matrix(loanj[,3:78])%>%t();dim(loanj1)<-c(76*33,1)
#
a<-cs1[rep(1:nrow(cs1),each=76),];a1<-data.frame(Trdmnt=rep(seq(as.Date("2000/3/1"),as.Date("2018/12/1"),"quarter"),33))
a2<-data.frame(a,Trdmnt=as.numeric(str_sub(str_replace_all(a1$Trdmnt,"-",""),start=1L,end=6L)))
a3<-data.frame(a2,badj=as.numeric(badj1),roaj=as.numeric(roaj1),capj=as.numeric(capj1),roej=as.numeric(roej1),
assetj=as.numeric(assetj1),equityj=as.numeric(equityj1),incomej=as.numeric(incomej1),
netimj=as.numeric(netimj1),servicej=as.numeric(servicej1),investj=as.numeric(investj1),
changej=as.numeric(changej1),hdj=as.numeric(hdj1),otherj=as.numeric(otherj1),
costj=as.numeric(costj1),netj=as.numeric(netj1),depositj=as.numeric(depositj1),loanj=as.numeric(loanj1))
# 计算出资本金率
cappj<-select(a3,1:3,equityj,assetj)%>%mutate(cappj=equityj/assetj*100)%>%select(-2,-4,-5)
#
a31<-mutate(a3,loa=loanj/assetj*100,gap=(loanj-depositj)/assetj*100,noint=(incomej-netimj)/incomej*100,
ser=servicej/incomej*100,exc=hdj/incomej*100,inv=investj/incomej*100,oth=(changej+otherj)/incomej*100,
cos=costj/incomej*100)%>%select(-(9:18))
a32<-arrange(a31,Trdmnt)%>%group_by(Stkcd)%>%slice(-1)%>%tbl_df()
a33<-arrange(a31,Trdmnt)%>%group_by(Stkcd)%>%slice(-n())%>%tbl_df()
a34<-data.frame(a32,asset1j=a33$assetj,deposit1j=a33$depositj,loan1j=a33$loanj)
a35<-mutate(a34,assetgrow=(assetj-asset1j)/asset1j*100,depgro=(depositj-deposit1j)/deposit1j*100,
            lngro=(loanj-loan1j)/loan1j*100)%>%select(-(9:10),-(19:21))%>%arrange(Stkcd,Trdmnt)%>%left_join(cappj,by=c("Stkcd","Trdmnt"))
a4<-rep(seq(as.Date("2000/7/1"),as.Date("2019/1/1"),"quarter"),33)
a5<-data.frame(a35,t=as.numeric(str_sub(str_replace_all(a4,"-",""),start=1L,end=6L)))%>%select(-Trdmnt) 
tr<-do.call("rbind",d);cr<-do.call("rbind",d1);br<-do.call("rbind",d2);pr<-do.call("rbind",d11);dr<-do.call("rbind",d12)
tr1<-filter(tr,tβ>0);cr1<-filter(cr,crisk>0);br1<-filter(br,brisk>0);pr1<-filter(pr,p>0);dr1<-filter(dr,d>0)
ris1<-full_join(tr1,cr1,by=c("Stkcd","Trdmnt"))%>%full_join(br1,by=c("Stkcd","Trdmnt"))%>%full_join(pr1,by=c("Stkcd","Trdmnt"))%>%full_join(dr1,by=c("Stkcd","Trdmnt"))
ris2<-left_join(ris1,a5,by=c("Stkcd","Trdmnt"="t"))%>%arrange(Stkcd)%>%select(-name)%>%select(-roej,-depgro,-lngro)
ris3<-filter(ris2,cappj!="NA",roaj!="NA",loa!="NA",gap!="NA",noint!="NA",cos!="NA",badj!="NA",
             assetgrow!="NA",ser!="NA",exc!="NA",inv!="NA",oth!="NA") # a<-mutate(ris3,aa=ser+exc+inv+oth)

# 回归 1
res<-lm(log(assetj)~cappj+roaj+loa+gap+noint+cos+badj+assetgrow,ris3)$residuals
# 通过固定效应求残差：res<-plm(log(assetj)~cappj+roaj+loa+gap+noint+cos+badj+assetgrow,model = "within",effect = "time",data = ris3, index = c("Stkcd","Trdmnt"))$residuals
ris4<-data.frame(ris3,rsize=res)
result<-plm(log(tβ) ~ rsize+cappj+roaj+loa+gap+noint+cos+badj+assetgrow,model = "within",effect = "time",
            data = ris4, index = c("Stkcd","Trdmnt"))
summary(result)
m<-as.matrix(summary(result)$coefficients)[,c(1,3,4)]
write.csv(m,"m.csv")
##直接回归 summary(lm(log(tβ) ~ rsize+capj+roaj+loa+gap+noint+cos+badj+assetgrow,ris4));summary(lm(log(tβ) ~ rsize+capj+roaj+loa+gap+ser+cos+badj+assetgrow,ris4))

# 回归 2 （分解非利息收入）
res<-lm(log(assetj)~cappj+roaj+loa+gap+ser+exc+inv+oth+cos+badj+assetgrow,ris3)$residuals
ris5<-data.frame(ris3,rsize=res)
result<-plm(log(brisk) ~ rsize+cappj+roaj+loa+gap+ser+exc+inv+oth+cos+badj+assetgrow,model = "within",effect = "time",
            data = ris5, index = c("Stkcd","Trdmnt"))
summary(result)
m<-as.matrix(summary(result)$coefficients)[,c(1,3,4)]
write.csv(m,"m.csv")
# 回归3 加入全部控制变量
result<-plm(log(tβ) ~ cap+roa+loa+gap+noint+cos+badln+assetgrow+equitytodebt+retainedearn+assetturn+lngro+depgro+sub+dou+los,model = "within",effect = "time",
            data = ris3, index = c("Stkcd","Trdmnt"))
summary(result)

# 描述统计分析：
ms1<-select(ris3,-1,-2)%>%mutate(ln=log(assetj));summary(ms1);apply(ms1,2,sd)

稳健性检验

# 稳健性检验：
## 1、提前1个季度的值
a4<-rep(seq(as.Date("2000/10/1"),as.Date("2019/4/1"),"quarter"),33)
##  提前2个季度的值
a4<-rep(seq(as.Date("2001/1/1"),as.Date("2019/7/1"),"quarter"),33)

## 2、固定银行效应
result<-plm(log(tβ) ~ rsize+cappj+roaj+loa+gap+noint+cos+badj+assetgrow,model = "within",effect = "individual",
            data = ris4, index = c("Stkcd","Trdmnt"))
summary(result)
m<-as.matrix(summary(result)$coefficients)[,c(1,3,4)]
write.csv(m,"m.csv")

## 双向固定效应
result<-plm(log(tβ) ~ rsize+cappj+roaj+loa+gap+noint+cos+badj+assetgrow,model = "within",effect = "twoway",
            data = ris4, index = c("Stkcd","Trdmnt"))
summary(result)
m<-as.matrix(summary(result)$coefficients)[,c(1,3,4)]
write.csv(m,"m.csv")

## 3、大小公司 分界点：2.085e+12 元
###小公司 1.945e+12
ris3<-arrange(ris3,assetj)%>%slice(1:(n()/2))
###大公司 
ris3<-arrange(ris3,assetj)%>%slice((n()/2+1):n())

## 4、 使用沪深300指数
findex<-read_excel("index.xlsx",sheet = 1)%>%select(2,3,12)%>%rename(Stkcd="简称",Clsdt="日期",Dretwd="涨跌幅(%)")%>%filter(Dretwd!="--")
findex1<-data.frame(findex[,c(1,2)],Dretwd=as.numeric(findex$Dretwd))
findex1<-(data.frame(findex1,Trdmnt=as.numeric(str_replace_all(findex1$Clsdt,"-","")))%>%select(-2))[,c(1,3,2)]
cs<-filter(findex1,Stkcd=="沪深300")%>%mutate(Cdretwdos=Dretwd/100)%>%select(-1,-3)

## 5、 控制传统风险衡量
res<-lm(log(assetj)~log(p)+log(d)+cappj+roaj+loa+gap+noint+cos+badj+assetgrow,ris3)$residuals
ris4<-data.frame(ris3,rsize=res)
result<-plm(log(brisk) ~ log(p)+log(d)+rsize+cappj+roaj+loa+gap+noint+cos+badj+assetgrow,model = "within",effect = "time",
            data = ris4, index = c("Stkcd","Trdmnt"))
summary(result)
m<-as.matrix(summary(result)$coefficients)[,c(1,3,4)]
write.csv(m,"m.csv")

求出各个银行的系统性尾部风险

tt<-group_by(tr,Stkcd)%>%summarise(t=mean(tβ))
cc<-group_by(cr,Stkcd)%>%summarise(c=mean(crisk))
bb<-group_by(br,Stkcd)%>%summarise(b=mean(brisk))
write.csv(tt,"tt.csv",row.names = FALSE);write.csv(cc,"cc.csv",row.names = FALSE);write.csv(bb,"bb.csv",row.names = FALSE)

上市时间	银行名称	系统性尾部风险	传染性风险	银行尾部风险
1991-04-03	平安银行	1.034	0.866	1.195
2007-07-19	宁波银行	1.005	0.834	1.204
1999-11-10	浦发银行	1.003	0.872	1.149
2003-09-12	华夏银行	1.046	0.859	1.215
2000-12-19	民生银行	0.896	0.846	1.061
2002-04-09	招商银行	0.900	0.871	1.032
2007-07-19	南京银行	0.945	0.840	1.124
2007-02-05	兴业银行	0.994	0.849	1.170
2007-09-19	北京银行	0.923	0.838	1.101
2010-07-15	农业银行	0.601	0.784	0.767
2007-05-15	交通银行	0.796	0.836	0.951
2006-10-27	工商银行	0.604	0.800	0.755
2010-08-18	光大银行	0.908	0.831	1.093
2007-09-25	建设银行	0.668	0.804	0.832
2006-07-05	中国银行	0.619	0.803	0.774
2007-04-27	中信银行	0.955	0.820	1.167

Jarry1

关注

1
点赞
踩
0

收藏

觉得还不错? 一键收藏
0
评论
对我国16家上市银行破产风险特征因素研究

#季度数据##银行特征与系统性风险library("readr")library("readxl")library("dplyr")library(stringr)library(PerformanceAnalytics)library(plm)setwd("e:/R/tail risk/day/finance")rf1<-read.csv("rf1.csv")listf<-read_excel("listf.xlsx",sheet = 1)%>%select(-2)
复制链接

扫一扫