smooth rolling regression coefficients
Posted: Wed Feb 08, 2017 1:57 am
Dear Gareth,
I got very bumpy coefficient series from rolling regressions. I learned that it is possible to smooth the estimates by fitting a kernel around the observations, giving the biggest weight to the observation in the centre and less weight to observations further away from the centre. My questions are:
1. Do you have other suggestions for smoothing the rolling estimates?
2. If the kernel is the solution, how can I do it in Eviews? I understand in Stata, you can achieve it by weighted least squares. What approach can I take in Eviews? Do you have some related sample codes?
Best,
Chang
I got very bumpy coefficient series from rolling regressions. I learned that it is possible to smooth the estimates by fitting a kernel around the observations, giving the biggest weight to the observation in the centre and less weight to observations further away from the centre. My questions are:
1. Do you have other suggestions for smoothing the rolling estimates?
2. If the kernel is the solution, how can I do it in Eviews? I understand in Stata, you can achieve it by weighted least squares. What approach can I take in Eviews? Do you have some related sample codes?
Best,
Chang