胖胖小龟宝 发表于 2018-1-5 15:12 
额……不能说很大,但我觉得5阶会比较好点
个人意见 仅供参考
您好,请问我在检验残差序列是否为白噪声的时候,能否帮我看下操作是否正确。操作如下X1,X2,X3,X4,X5 open as var——lag interval 默认为2——view——residual test——Autocorrelation LM test——lag to include填了12——结果如下:
VAR Residual Serial Correlation LM Tests
Null Hypothesis: no serial correlation at lag order h
Date: 01/05/18 Time: 16:23
Sample: 12/28/2001 12/26/2017
Included observations: 3790
Lags LM-Stat Prob
1 482.1574 0.0000
2 404.1346 0.0000
3 612.2835 0.0000
4 96.87586 0.0000
5 111.5663 0.0000
6 88.33782 0.0000
7 52.86352 0.0009
8 60.96764 0.0001
9 79.76931 0.0000
10 64.38263 0.0000
11 36.23971 0.0680
12 89.93549 0.0000
Probs from chi-square with 25 df.