I am planning to analyse the impact of two types of ads on consumer emotions (measured with AI facial emotion recognition while exposed to advertising)
I am using a Panel VAR model that includes exogenous variables ( two dummy variables (1, 0) indicating each type of ads), and I am interested in assessing the impact of this exogenous variable on the endogenous variables (happy, sad, .....) .
I went to use the panel VAR followed by IRFs (impulse response functions) to assess this impact, but as far as I know, IRFs are used just to assess the impact of an endogenous variable on other endogenous variables.
However, my questions are:
Correct, IRFs are typically for endogenous variables, not exogenous.
For exogenous effects: Use generalized impulse response functions (GIRFs) or structural VAR (SVAR) models where you treat your dummies as shocks.
Software: In R, you can use the pvar package for panel VAR and vars or plm for SVAR analysis.
Alternative method: Consider using dynamic panel data models (e.g., GMM) if panel VAR isn't appropriate.
- Badges
- Science topic