Which lagged variable to include? (Wald Tests)

[Andreas Baltin]

Hi Statalisters,

I am currently running a FE regression for a 2-period panel data set in China. I have a variable on homicides by state. There are 3 versions of this variable. One is lagged by 1 month, one lagged by 6 months, one lagged by 12 months, and I just don't know which one to use in my regression.

This paper states: "Based on Wald tests for model specification, including six months of lagged homicide rates was preferred to including one month and three months of lagged homicide rates",
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5076372/

However, they do not say how they carried out these tests/ how exactly they came to this conclusion. Is someone able to provide some guidelines/code showing how exactly I do this?

Many thanks for your help!
Andreas

[Andreas Baltin]

Does anyone have an idea?

[Phil Bromiley]

You'll increase your chances of a useful answer by following the FAQ on asking questions. After posting 25 questions this month, I think you should know this.

How to choose lags is somewhat unclear. I think some use AIC or BIC. I suppose you could include all three in the model and keep the statistically significant one. Have you looked at this in an econometrics text?