Page 1 of 1

smooth rolling regression coefficients

Posted: Wed Feb 08, 2017 2:24 pm
by shuchang
Dear Sir/Madam,

I got bumpy coefficient series when I run rolling regressions (a fixed effect model in a pool object). I learn that it can be smoothed by fitting a kernel around the centre of the observations, giving the observations in the centre the biggest weight, and less weight to the observations further away from the central observations. People I know have coded this in Stata.

My questions are: Can this be achieved in Eviews? Are there alternative ways of smoothing rolling regression coefficient series?

Best,

Chang