Inclusion of AR terms in Logit model
Posted: Thu Nov 17, 2016 6:02 am
Hi to all,
I´m from Brazil and at the moment I am trying to model the market share of ethanol, I think the best way to do this is by using a Logit model like the one below:
Ethanol_share = 1/(1+exp(c(1)+c(2)*Ethanol_price/Gasoline_Price))
The problem that I am facing is that there is serial correlation in the model above, so I included an AR term to correct for the serial correlation, the model became like this.
Ethanol_share = 1/(1+exp(c(1)+c(2)*Ethanol_price/Gasoline_Price)) + [AR(1)=c(3)]
By including the AR term the serial correlation problem was corrected.
My question is if it is wrong to include AR terms in Logit models?
I´m from Brazil and at the moment I am trying to model the market share of ethanol, I think the best way to do this is by using a Logit model like the one below:
Ethanol_share = 1/(1+exp(c(1)+c(2)*Ethanol_price/Gasoline_Price))
The problem that I am facing is that there is serial correlation in the model above, so I included an AR term to correct for the serial correlation, the model became like this.
Ethanol_share = 1/(1+exp(c(1)+c(2)*Ethanol_price/Gasoline_Price)) + [AR(1)=c(3)]
By including the AR term the serial correlation problem was corrected.
My question is if it is wrong to include AR terms in Logit models?