Hello community! I have a quick question.
-Dummy variable takes value of 1 if developed, takes value of 0 if developing
-Dummy variable takes value of 1 for post-crisis (2009-2015), takes value of 0 for pre-crisis (2005-2008)
My question is:
Can I use the interaction between them, developed*post, and interact with my explanatory variables?
Example:
Dependent variable: GDPgrowth
Independent Variables: Infrastructure, Capital Formation, Labor Force, Trade
Can I do the following?:
GDPgrowth= Infrastructure + Capital Formation + Labor Force + Trade + Infrastructure*developedpost + Capital Formation*developedpost + Labor Force*developedpost + Trade*developedpost
What I want to achieve is for example:
Is the impact of infrastructure on GDP growth significant in developed countries after the crisis, in comparison with developing countries before crisis.
(This is a panel data).
Thank you very very much!
I can't find this anywhere..
Yes, you can.
See this link:
https://www.researchgate.net/post/Is-an-interaction-between-two-dummy-variables-possible-If-so-what-model-can-I-use-to-estimate-the-effect-of-the-interaction-on-the-y-variable
yes, ,but I think you should introduce a treeway interaction to estimate the full set of coefficients, for axample, interaction a*b*c would estimate a, b, c, ab, ac, cb, abc. In your strategy instead you are assuming that the baseline coefficient is pre-crisis but putting together developed or not developed.
best,
P.E.
Piero Esposito
Dear Pierro,
Let me start by thanking you for your answer.
Have some questions regarding your answer:
When you mention baseline coefficient, you mean all the coefficients of the independent variables, or just the original variables that were interacted with the dummies terms?
best,
H.S.
Dear Enrique,
yes, baseline coefficient means the non-interacted terms. In your case, you have a continous variable interacted with developed*post crisis. So the interaction you propose would tell you the diffential effect for developed countries in the post crisis period. So this means that all other combinations (develped pre-crisis, non-developed pre and post crisis) are pooled together. If you have a justification for that, it's fine. Otherwise you sould interact, for example, capital formation (K) with both dummies (crisis and developed). In this way, the baseline category would be developing countries-pre crisis and you could get the differential effect of 1) the crisis (the coefficients for K*crisis), being a developed country (K*developed) and being developed and in post crisis (K*developed*crisis).
I hope this was clear,
best,
P.E.
- Badges
- Science topic
- Similar topics
- Geoscience
- Asia
- India