Nelson siegel model estimed by Kalman Filter

Question

Alberto 2012-9-6

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/47523-nelson-siegel-model-estimed-by-kalman-filter

回答： asma noor 2019-7-3

Hi, I'm having some trouble in estimating the Nelson Siegel model with the Kalman Filter according to the metodology presented in the paper "The macroeconomy and the yield curve: a dynamic latent factor approach". Can I share some ideas with someone who is familiar with this approach.

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Horace 2013-12-1

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/47523-nelson-siegel-model-estimed-by-kalman-filter#answer_116851

Hi, I hope you're well.I'm working with the Nelson Siegel model as well.I'm wondering whether you have solved it and how you solved it.Thank you!

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Alberto 2014-7-15

I was able to get convergence only with the exclusion of macroeconomic variables from the state equation. So in order to asses the link between state variable and macro I needed to estimante another model.

请先登录，再进行评论。

Answer 2

Jonas Striaukas 2014-7-15

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/47523-nelson-siegel-model-estimed-by-kalman-filter#answer_145130

I have a code, but for me it does not converge when I have Q_t (3x3 measurement error matrix) positive definite and it does if I have diagonal matrix. Maybe someone had the same issue? i think is to do with initializing Kalman filter but not sure..

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Alberto 2014-7-15

my personal experience is that starting values are a very tricky issue with the kalman filter. If you are trying to replicate a paper you may use the final estimed parameters of the paper as starting values for your problem.

请先登录，再进行评论。

Answer 3

asma noor 2019-7-3

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/47523-nelson-siegel-model-estimed-by-kalman-filter#answer_381832

i am also working with nelson siegle model estimated by kalman filter but its complex one anyone there who help me out how i can estimate that command.