[求助] 用GAUSS来做markov-switching GARCH的code

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

236856.rar (11.37 KB)

2008-8-15 07:11:00 上传

[求助] 用GAUSS来做markov-switching GARCH的code

论文想用markov-switching GARCH model，就是和Hamilton(1994) 'Autoregressive conditional heteroskedasticity and changes in regime" 同样的模型。

在网上找到了Hamilton上传他这个paper的数据和gauss code，可是我做不出结果。

一开始我的GAUSS没有Optimal的模块，然后在另个帖子下载了这个模块后，说所有optmum.ext里的code是undefined。。

不知哪位高手能帮忙看看，指点迷津。。。数据和code我已经上传了。。。

谢谢，谢谢，非常感谢！！！！！

扫码加我 拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

分享0 收藏6 回帖

关键词：switching Markov switch GARCH GAUSS GARCH GAUSS code

rder of autoregression       1.0000000
Order of ARCH process       2.0000000
Number of primitive states       3.0000000
Number of lagged states that affect y       2.0000000
First observation used for estimation is       5.0000000
with leverage effect
distribution is t
Constant term in regression      0.35080500
Autoregressive coefficients in regression      0.25003200
Initial variance not neeeded
Constant term in ARCH process      0.56817500
Coefficients on lagged epsilon squared in ARCH process     0.027988000       0.11706300
(Transposed) matrix of transition probabilities
      0.99236621       0.00000000     0.0025702358
    0.0076337928       0.99144538      0.014353048
      0.00000000     0.0085546244       0.98307672

The state with no adjustment to ARCH process is state 1, with transition
probability       0.99236621
Vector of variance factors for states 2 through       3.0000000        4.3512660        13.146594
Coefficient on negative lagged change for asymmetric effect      0.42415600
degree of freedom for t distribution is       7.2199370

Initial values:      0.35080500       0.25003200      -0.56817500      0.027988000      -0.11706300        11.401600      0.051132000        10.765493       0.12083100        4.3512660        13.146594       0.42415600       -5.2199370
Initial value for negative log likelihood:       2802.6510

Do you wish to continue (y or n)?

[此贴子已经被作者于2008-8-22 11:48:27编辑过]

======================================================
          FINAL ESTIMATES

Value of log likelihood:      -2802.6510

Coefficients:
      0.35080500       0.25003200      -0.56817500      0.027988000      -0.11706300        11.401600      0.051132000        10.765493       0.12083100        4.3512660        13.146594       0.42415600       -5.2199370

Constant term in regression      0.35080500
Autoregressive coefficients in regression      0.25003200
Initial variance not neeeded
Constant term in ARCH process      0.56817500
Coefficients on lagged epsilon squared in ARCH process     0.027988000       0.11706300
(Transposed) matrix of transition probabilities
      0.99236621       0.00000000     0.0025702358
    0.0076337928       0.99144538      0.014353048
      0.00000000     0.0085546244       0.98307672

The state with no adjustment to ARCH process is state 1, with transition
probability       0.99236621
Vector of variance factors for states 2 through       3.0000000        4.3512660        13.146594
Coefficient on negative lagged change for asymmetric effect      0.42415600
degree of freedom for t distribution is       7.2199370

Gradient vector:
   -0.0018811481    -0.0055793042     0.0016680838    -0.0030521296     0.0022067962   1.3614388e-005    -0.0043252972  -7.9209387e-006    -0.0034751487   -0.00014573237   2.6884889e-005   -0.00029098294   4.0814785e-005
Standard errors:
     0.048607937      0.028576001       0.11344921      0.040766654      0.046181942        7.7422944      0.024049265        3.9032090      0.048562656       0.95006788        3.2393254      0.098623965        1.3745019

Matrix of Markov transition probabilities:

      0.99236621       0.00000000     0.0025702358
    0.0076337928       0.99144538      0.014353048
      0.00000000     0.0085546244       0.98307672

Ergodic probs for full state vector:
      0.10002282    0.00076942716       0.00000000   7.1054274e-015    0.00076871320   6.6327937e-006   7.1054274e-015       0.00000000   5.6843419e-014   1.4210855e-013   7.0499162e-015       0.00000000   2.1316282e-014       0.58660244     0.0050614624   1.3121401e-005   7.3274249e-005     0.0050187393    0.00076942716   5.9188306e-006       0.00000000   1.4210855e-014     0.0042927492   3.7039718e-005    0.00076222459     0.0042565147       0.29153949

Ergodic probs for primitive states:
      0.10156760       0.59676903       0.30166337

Log likelihood:
      -2802.6510
Probabilities for primitive states
filtered probabilities
Obs return  P(st-0 =1 ) P(st-0 =2 ) P(st-1 =1 ) P(st-1 =2 ) P(st-2 =1 ) P(st-2 =2 )

5.0000 2.0848 0.0757 0.6142 0.0764 0.6142 0.0766 0.6142
6.0000 -0.7567 0.0493 0.6515 0.0496 0.6512 0.0504 0.6507
7.0000 1.1276 0.0630 0.6995 0.0623 0.6980 0.0625 0.6961
8.0000 2.3167 0.0345 0.7298 0.0348 0.7316 0.0348 0.7298
9.0000 0.5900 0.0646 0.7814 0.0637 0.7814 0.0638 0.7808
10.0000 0.0975 0.1052 0.7949 0.1049 0.7962 0.1028 0.7945
11.0000 -0.7480 0.0867 0.8316 0.0873 0.8345 0.0871 0.8347
12.0000 0.2860 0.1368 0.8105 0.1368 0.8127 0.1373 0.8137
13.0000 -3.8304 0.0056 0.7684 0.0069 0.7937 0.0076 0.8075
14.0000 -1.9898 0.0068 0.7911 0.0062 0.7912 0.0076 0.8136
15.0000 1.2164 0.0080 0.8190 0.0077 0.8205 0.0069 0.8191
16.0000 -0.7302 0.0058 0.8432 0.0056 0.8465 0.0055 0.8471
17.0000 -7.4329 0.0003 0.4249 0.0004 0.4625 0.0005 0.4839
18.0000 4.1470 0.0004 0.4214 0.0003 0.4201 0.0004 0.458

[此贴子已经被作者于2008-8-22 11:49:23编辑过]

xuehe ,本人之前从未接触过GAUSS软件,问一个非常菜鸟的问题:

在下载了楼主的MARKOV SWITCHING的估计模块和数据后,怎样用模块来对数据进行分析.具体步骤是什么样的啊?

我在命令窗口通过FILE-OPEN 打开模块,然后直接RUN-RUN MAIN FILE,结果说找不到optmum.icg.什么是optmum.icg呀?  

先谢了!

[此贴子已经被作者于2008-8-24 1:10:06编辑过]

可能你没装optmum或optimal的toolbox,网上有下的。

xuehe,你好

下了hamilton的模型的数据和CODE,

但是运行的时候总是出现

>>run F:\Hamilton 1994 Markov switching ARCH.4\Maxseek;
Line 53 in F:\Hamilton 1994 Markov switching ARCH.4\maxseek
   Syntax error G0008 : 'proc'
Line 56 in F:\Hamilton 1994 Markov switching ARCH.4\maxseek
   WARNING: local outside of procedure G0017 : 'local th'
Line 66 in F:\Hamilton 1994 Markov switching ARCH.4\maxseek
   retp outside of procedure G0055

能帮我看下这是怎么回事么?

还有下了CODE文件夹后,应该放在哪个目录下?GAUSS界面的上的文件路径该怎么设置?

期待解答!

放在GAUSS下，具体操作见邮件。

版主，可否把这个邮件也给我发一份，多谢！

学习一下~，有点难

邮件能否给我一份，谢谢。yancheng54@163.com
1# jjjjdddd