[问答] 请问如何从这个ECM结果看出因果关系 [推广有奖]

0关注
0粉丝

小学生

71%

还不是VIP/贵宾

-

0%

威望: 0 级
论坛币: 10 个
通用积分: 0
学术水平: 0 点
热心指数: 0 点
信用等级: 0 点
经验: 53 点
帖子: 3
精华: 0
在线时间: 11 小时
注册时间: 2016-1-13
最后登录: 2016-8-14

楼主

hanyu_0820

发表于 2016-8-2 19:58:03 |AI写论文

是否 +2 论坛币

k人参与回答

经管之家送您一份

应届毕业生专属福利!

求职就业群

赵安豆老师微信：zhaoandou666

经管之家联合CDA

送您一个全额奖学金名额~ !

立即领取

感谢您参与论坛问题回答

经管之家送您两个论坛币！

+2 论坛币

论文要研究欧洲政府债券收益GBY和信用违约互换CDS的联动关系，老师建议我用四个公式，我对公式目的不太懂，望大神解答！
1.Gby(t ) = constatnt + b* cds (t ) + error term
这个公式是协整回归，应该是确定GBY和CDS有没有长期的关系
2.Dgby (t ) = constant + b1 Dcds (t ) + b2 u(t-1) + error
这个公式是描述在公式1中长期均衡关系下的短期调整吧，但是我不太懂为什么
3.Dgby(t ) = constant + b1*Dgby (t -1) + b2*Dgby (t-2) +c1*Dcds (t-1) + c2*Dcds (t-2) + d*u(t-1) + error term
4.Dcds(t ) = constant + b1*Dgby (t -1) + b2*Dgby (t-2) +c1*Dcds (t-1) + c2*Dcds (t-2) + d*u(t-1) + error term
这两步ECM 模型是通过系数是否为零来确定GBY和CDS到底哪个影响另一个，
如果式3中c1=c2=d=0.式4中b1 b2 d≠0，则GBY 影响CDS
如果式3中c1 c2 d不为0，式4中b1=b2=d=0,则CDS影响GBY
如果3中c1 c2 d和4中b1 b2 d至少一个不为0，则互相影响
以奥地利为例我的结果如下图，求大神分析
式3：
Dependent Variable: DAUSGB
Method: Least Squares
Date: 08/01/16 Time: 22:15
Sample (adjusted): 2008M10 2016M04
Included observations: 91 after adjustments
HAC standard errors & covariance (Bartlett kernel, Newey-West fixed
      bandwidth = 4.0000)

Variable Coefficient Std. Error t-Statistic Prob.

C -0.046187 0.026941 -1.714370 0.0901
DAUSGB(-1) -0.036804 0.124622 -0.295328 0.7685
DAUSGB(-2) 0.092733 0.117391 0.789945 0.4318
DAUSCDS(-1) -0.001837 0.001066 -1.722776 0.0886
DAUSCDS(-2) 0.000939 0.001389 0.676452 0.5006
AUSU(-1) -0.028571 0.020170 -1.416530 0.1603

R-squared 0.074259       Mean dependent var -0.047615
Adjusted R-squared 0.019803       S.D. dependent var 0.210904
S.E. of regression 0.208805       Akaike info criterion -0.231172
Sum squared resid 3.705958       Schwarz criterion -0.065621
Log likelihood 16.51835       Hannan-Quinn criter. -0.164383
F-statistic 1.363664       Durbin-Watson stat 2.025128
Prob(F-statistic) 0.246128       Wald F-statistic 1.705606
Prob(Wald F-statistic) 0.142063

式4：

Dependent Variable: DAUSCDS
Method: Least Squares
Date: 08/01/16 Time: 22:16
Sample (adjusted): 2008M10 2016M04
Included observations: 91 after adjustments
HAC standard errors & covariance (Bartlett kernel, Newey-West fixed
      bandwidth = 4.0000)

Variable Coefficient Std. Error t-Statistic Prob.

C -0.668036 2.282844 -0.292633 0.7705
DAUSGB(-1) -4.377261 6.818874 -0.641933 0.5226
DAUSGB(-2) -17.72755 10.09198 -1.756597 0.0826
DAUSCDS(-1) 0.077958 0.143378 0.543725 0.5881
DAUSCDS(-2) 0.085529 0.149016 0.573959 0.5675
AUSU(-1) 6.535706 2.835298 2.305121 0.0236

R-squared 0.133560       Mean dependent var 0.113330
Adjusted R-squared 0.082593       S.D. dependent var 21.43995
S.E. of regression 20.53547       Akaike info criterion 8.945844
Sum squared resid 35844.98       Schwarz criterion 9.111396
Log likelihood -401.0359       Hannan-Quinn criter. 9.012634
F-statistic 2.620521       Durbin-Watson stat 2.023137
Prob(F-statistic) 0.029737       Wald F-statistic 2.577458
Prob(Wald F-statistic) 0.032060