EViews.com

Hello again everybody,

a very simple question at first. Is it sufficient to use "cointreg" In order to estimate a simple error correction model (not VEC!)? If not, how do I specify and estimate an error correction model in Eviews (not VEC)?

Thank you very much,

Chris




Hello everybody,

I am using Eviews 7 and currently I am trying to implement an error correction model and have some questions regarding estimation of the model.

Lets assume the model looks like this (long-term relation):
Price(t)=beta0+beta1*Reg_1(t)+beta2*Reg_2(t)+error(t)

Then, the error correction model should look like (short-term relation):
d(Price(t))=b1*d(Reg_1(t))+b2*d(Reg_2(t)+lambda*(Price(t-1)-beta0-beta1*Reg_1(t-1)-beta2*Reg_2(t-1))+error_u(t)

In order to estimate the model, I am using DOLS:
equation f_ecm.cointreg(method=dols, trend=const, regtrend=none, lltype=aic, cov=hac) Price Reg_1 Reg_2

Question 1:
After estimation of the model I checked "representations", which says estimation equation is Price(t)=C(1)+C(2)*Reg_1+C(3)*Reg_3. I would like to know whether the error correction model is specified correctly or did I miss any restrictions or anything else?

Question 2:
Lets assume the model is to be estimated based on OLS. Would the specification listed below be correct?
equation f_ecm2.ls(cov=hac) d(Price)=c(1)*d(Reg_1)+c(2)*d(Reg_2)+c(3)*(Price(-1)-c(4)-c(5)*Reg_1(-1)-(6)*Reg_2(-1))

Thank you very much for your support and best regards.

Chris