本人在做VAR模型滞后期选择的时候,在lags to include 填的数据越大(从1到5),滞后期就应该选谁,当我填6的时候就现实log of non positive number 这是怎么回事呢?我的数据全为正数,一般是带*多的在中间才确定最佳滞后阶数。再次附上我的数据,请好心人帮我看看,对我论文很重要。后面还要做协整检验,脉冲响应,方差分解。非常感谢
small sample numbers hard to do empirical studies, be careful of small sample bias
basically, choose the var lag, depending on the maximum var lag of your data type; ex. annual, quarterly, monthly, and so on...