How to test the difference of estimated coefficient from two models

[Ken Qi]

Hello, I am trying to run two interaction models and test the two estimated coefficients differences.

For example, my model 1 is xtreg DV (controls) L1.c.(iv)##L1.c.(moderator). My model 2 is xtreg DV (controls) L1.c.(iv)##L3.c.(moderator) trying to explore whether the 3-year lagged value of the moderator has a stronger moderation effect, how should I do it to test it?

If I shouldn't do this way, please kindly advise what I should do instead. Thank you very much!

Ken

[Carlo Lazzaro]

Ken:
welcome to this forum.
See Comparing coefficients across two models (same X data, but slightly different Ys) - Statalist

[Ken Qi]

Thanks, Carlo. I looked at that post, but feel my situation is a little different.

My model 1 is xtreg DV(controls) L1.c.xrd##L1.c.TOTAL_IT_BUDGET_GVKEY_PER_EMPL, where moderator is 1-year lagged IT investment var, and second model is xtreg DV (controls) L1.c.xrd##L3.c.TOTAL_IT_BUDGET_GVKEY_PER_EMPL, where moderator is a 3-year lagged value of the same IT investment var. I tried to follow Clyde's suggestion of using regress instead of xtreg and running suest and using
test [Y1_mean = Y2_mean]: X1, but I am not sure what to put here as X1,should that be the interaction b/w xrd and TOTAL_IT_BUDGET_GVKEY_PER_EMPL or just TOTAL_IT_BUDGET_GVKEY_PER_EMPL? I would appreciate any more guidance. Thank you. I am new to Stata. Ken

[Bruce Weaver]

Hello Ken. I'm not (currently) posting to suggest a solution. I'm just trying to make your two models more comparable for other members by using code delimiters (the # key on the toolbar) and lining them up vertically.

xtreg DV (controls) L1.c.xrd##L1.c.TOTAL_IT_BUDGET_GVKEY_PER_EMPL // Model 1
xtreg DV (controls) L1.c.xrd##L3.c.TOTAL_IT_BUDGET_GVKEY_PER_EMPL // Model 2

* Model 1: moderator is 1-year lagged IT investment var
* Model 2: moderator is a 3-year lagged value of the same IT investment var.

[Ken Qi]

Okay, thanks, Bruce!

[Carlo Lazzaro]

Ken:
you may want to take a look at Compare coefficients from two separate panel regressions in Stata - Cross Validated (stackexchange.com), that points the reader out to the community-contributed module -xtsur-.