[Gretl-users] question about loop-function in GRETL

Dear sir or madam, I'm a belgian student and this week I've been studying my course of Advanced Business Econometrics. I'm using the book ' Principles of Econometrics' (3rd version) from Hill, Griffiths & Lim. Now I am at Chapter 14: Time varying volatility and ARCH models. I got a problem with ex. 14.10 p. 381: this GARCH in mean model is written as a function of the time-varying standard deviation. The main problem is: I don't know how to write the time-varying standard deviation in the right form in the script. At the moment, my script seems like this (I thing the red piece is missing, but that's not enough I think): Estimate a T-GARCH-in-mean model and check that you obtain the following results: Ŷt = - 0,407 + 1,983 √ht (t) (-2,862) (5,243) Ht = 0,022 + (0,211 – 0,211dt-1)e²t-1 + 0,782 ht-1 (4,697) (8,952) (-8,728) (27,677) my script seems like this: function gim_filter(series y, \ scalar mu, scalar theta, scalar delta, scalar alpha, \ scalar gam, scalar beta, series *h)series lh = var(y)series le = y - muscalar T = $nobsloop for i=3..T --quiet scalar ilag = $i - 1 scalar d = (le[ilag]<0) scalar e2lag = le[ilag]^2 lh[i] = delta + alpha*e2lag + gam*e2lag*d + beta*lh[ilag] le[i] = le[i] - theta*lh[i]^(1/2)end loopseries h = lhreturn series leend functionopen c:\Temp\ukscalar mu = 0.8scalar gam = 0.1scalar alpha = 0.4scalar beta = 0scalar delta = 0.5scalar theta = 0.1series h = NAmle ll = -0.5*(log(2*pi) + log(h) +(e^2)/h) e = gim_filter(r, mu, theta, delta, alpha, gam, beta, &h) params mu theta delta alpha gam betaend mle --robust Would you be so kind to look after this exercise and let me know something?Yours faithfully, Maarten De nieuwe Hotmail maakt je leven nog makkelijker Wees er als eerste bij… _________________________________________________________________ Alle fun stuff van Messenger nu verzameld op één coole site! http://www.messengerbillboard.be/nl