Response Standard Errors - VAR, Monte Carlo
Posted: Wed Jan 20, 2010 2:50 pm
Hello alltogether!
I estimate the Impulse-Response of a VAR. For the standard error, I use the Monte Carlo procedure.
What is the distribution in connection with the standard error estimated by MC? I'd like to use different confidence intervals e.g. 95% or 90%. If the distribution would be normal, I would take 1,96*standard deviation for 95% or 1.68*sd for 90%.
If I display the Impulse Response, then EViews uses error bounds with 2*sd without saying to which confidence interval that matches. What is the reason?
I would be very obliged for information about the distribution, the MC standard error refers to.
Best regards,
Stefan
I estimate the Impulse-Response of a VAR. For the standard error, I use the Monte Carlo procedure.
What is the distribution in connection with the standard error estimated by MC? I'd like to use different confidence intervals e.g. 95% or 90%. If the distribution would be normal, I would take 1,96*standard deviation for 95% or 1.68*sd for 90%.
If I display the Impulse Response, then EViews uses error bounds with 2*sd without saying to which confidence interval that matches. What is the reason?
I would be very obliged for information about the distribution, the MC standard error refers to.
Best regards,
Stefan