regression with ARMA_GARCH errors
7 views (last 30 days)
Show older comments
I want to estimate a regression model with ARMA(3,3) - GARCH(1,1) errors:
y_t = bX_t + e_t
e_t = a_1e_{t-1}+...+a_3e_{t-3} + u_t + r_1u_{t-1}+...+r_3u_{t-3}
u_t = h_tε_t
h_t = ω + αu^2_{t-1} + βh_{t-1}
I know I can do it in two steps. Estimate the regression first and then fit an ARMA model with GARCH variance to the residuals of the regression, however I was wondering If I could do it in one step. The obvious choice would be a regARIMA model but regARIMA model doesn't support garch(1,1) value in 'Variance'.
Thank in advance
2 Comments
Giovambattista Perciaccante
on 20 Dec 2017
suppose your observed variable is called:
yt
specify the model for the conditional variance as GARCH(1,1)
varmdl = garch('GARCHLags', 1, 'ARCHLags', 1)
You haven't estimated anithing yet.
Then specify the model for the mean as ARMA(3,3). Inside the function ARIMA you call the model for the variance specifying the option 'Variance'
mdl = arima('ARLags',3, 'MALags',3,'Variance', varmdl)
Again you haven't estimated anything yet.
Finally using the function ESTIMATE you estimate all the coefficients in one go:
es = estimate(mdl, yt)
Answers (0)
See Also
Categories
Find more on Conditional Variance Models in Help Center and File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!