After estimating an ARDL model in eviews, you can ask for the long run coefficients by clicking view > coefficients diagnostic > long run form and bound test. My question is: how standard errors of the long run coefficients are calculated? I know they are related to "Delta method" and I tried to search more about it and how standard errors of the long run coefficients are calculated in eviews user guide, but I couldn't.
I would appreciate any guide in how to calculate standard errors of the long coefficients.
Thanks & regards.
Dear Nip Nip,
I am not quite familiar with the ARDL model and its long-run coefficients, but I imagine the procedure is as follows.
1. Write the analytical expressions for the estimators of the coefficients. This is a multivariate function of random variables (before you have seen your data). The expected value of this multivariate function gives the estimated long-run coefficients.
2. Use the Delta method to compute an approximate value of the variance of the multivariate function.
3. Estimate this variance, and take the square root to obtain estimated standard errors.
A good text on the Delta method (also called "The method of statistical differences") is Elandt-Johnson and Johnson "Survival models and data analysis" (Wiley 1980); see Section 3.14.2. No doubt, you will find other good explanations of this method on the Internet.
Good luck.
Nico
see for instance on the internet the stata explanation: http://www.stata.com/support/faqs/statistics/delta-method
- Badges
- Science topic
- Similar topics
- Mathematics
- Applied Mathematics