Dummy for Halloween-Effect

[Nip]

Hi,

I have to test the halloween-effect in Eviews. The halloween-effect means that the returns in the time november-april are higher than in the time from mai-oktober.

What is the right equation for this ?

My proposal is:

Return=c(1)+c(2)*(@month>=11 or @month<=4)

Is this correct ?

[EViews Gareth]

Yes, that will enter a dummy for Nov. - April.

[nic]

Hi,

I have to test the halloween-effect in Eviews. The halloween-effect means that the returns in the time november-april are higher than in the time from mai-oktober.

What is the right equation for this ?

My proposal is:

Return=c(1)+c(2)*(@month>=11 or @month<=4)

Is this correct ?

Yes. If C(2) is significantly positive, you will have Halloween effect. Alternatively, you can model without intercept and insert two dummies one for Nov-April and another for may to Oct.Keep in mind, Eviews do no generate F statistics in this case. This is the case of regression through origin.