regression (OLS) with lags.

[Andy Lee]

Hi everyone.
I am doing thesis about the causal relationship between A and B. It means : "does there have the relationship between A and B? if have, it from A to B or B to A".

1. Augmented Dickey-Fuller : all A and B variable is not have unit root test, means all two variables is STATIONARY right at I(0).

2. lag order preestimation: I choose max lags is 10. I choose SBIC criterion => the best K = 4

3. Run the VARgranger after run VAR. I find that the "A GRANGER-CAUSE B". and it is unidirectional. B does not granger cause A.

4. Here. I want to run the regression to know how the A granger cause B (positive or negative, how percentages? ). It means that I maybe run regression OLS with lags. But I do not know anythings, or step in STATA to do this. !!!

Please, help me in this problem. Please help me with the step by step in most detailed. tks so much. !!!!

[Andy Lee]

Anyone here?
please help me

Should I write in stata is: 1. regress B = L(1/4).B + L(1/4).A or 2. regress B = L(1/4).A or 3. regress B = A

[Phil Bromiley]

You didn't get a quick answer. You'll increase your chances of a useful reply by following the FAQ on asking questions - provide Stata code in code delimiters, Stata output, and sample data using dataex.

You seem to be doing fancy stuff when you don't see to know the basics. Also, if you have 3 options for writing an equation, try them out.
regress B L(1/4).B L(1/4).A
would probably run fine. Note, you do not use the equals sign or plus sign in a regress statement (look at the documentation or use the interactive menus!).

Whether you should include lagged B's or A's or neither is a substantive rather than a Stata issue. Some on the list might help you with granger causality. but I'm not the one.