Adjusting standard errors for generated regressors
Posted: Mon Jul 03, 2017 12:50 am
Hi everyone,
Following Hamilton "Why You Should Never Use the HP Filter" (2017) I am estimating output gap using a simple OLS regression on (log) Real GDP and taking the residuals as a regressor in a second-step regression.
Consider a first-step dynamic regression on x(t):
x(t+h) = c + x(t) + x(t-1)+ ... + x(t-p) + Espilon(t)
where h=8 and p=4 on quarterly data (Hamilton's advice). The residuals are then interpreted as the cyclical component of x(t).
Now, if one wants to use the residuals into a second step regression (let's say a Taylor rule or a fiscal rule), he will have to correct the variance-covariance matrix for the generated regressors problem in the second step.
Is there a way into Eviews to correct manually (by programming) this var-covar matrix after having estimated the second-step regression and re-display the results?
Thanks in advance for any help!
Following Hamilton "Why You Should Never Use the HP Filter" (2017) I am estimating output gap using a simple OLS regression on (log) Real GDP and taking the residuals as a regressor in a second-step regression.
Consider a first-step dynamic regression on x(t):
x(t+h) = c + x(t) + x(t-1)+ ... + x(t-p) + Espilon(t)
where h=8 and p=4 on quarterly data (Hamilton's advice). The residuals are then interpreted as the cyclical component of x(t).
Now, if one wants to use the residuals into a second step regression (let's say a Taylor rule or a fiscal rule), he will have to correct the variance-covariance matrix for the generated regressors problem in the second step.
Is there a way into Eviews to correct manually (by programming) this var-covar matrix after having estimated the second-step regression and re-display the results?
Thanks in advance for any help!