I am looking to calculate and explain abnormal returns (AR) on dividend announcement events:
Regression: R = a + B(MarktReturn) + e E(R) = a + BMktReturn
Therefore AR = R-ER
To explain the AR, I want to test few hypothesis so I do another regression with CAR (CAR=cumulative AR):
CAR = a + B1(Size) + B2(0-10%Increase) + B3(10-25% increase)+B4(25+ Increase)
- 0-10% increase is a dummy variable that takes 1 if the dividend increase amount was between 0-10%, 0 otherwise.
- 10-20%% increase is a dummy variable that takes 1 if the dividend increase amount was between 10-20%, 0 otherwise.
- Same logic for 25+ Increase.
Is my method coherent/valid with what I am looking to achieve?