coding Component ARCH model. Need URGENT help!

[narinka]

Hi Guys,

I am glad I found this forum, so hopefully I can get some help here. I need to replicate Adrian Rosenberg 2008 paper "Stock Returns and Volatility:
Pricing the Short-Run and Long-Run Components of Market Risk" for my term paper project , and I am struggling really badly with this (mainly because of lack of programming skills). I wrote to the author and he told me I can estimate the two factor model in Eviews. I just hope that someone can help me here. The model is the following:

1(a) RM(t+1)=mu(t)+sqrt(v(t))*ε(t+1)
1(b) ln(sqrt(v(t))=s(t)+l(t)
3) mu(t)=theta1+theta(2)*s(t)+theta3*l(t)
1(c) s(t+1)=theta4*s(t)+theta5*ε(t+1)+theta6*(abs(ε(t+1))-sqrt(2/pi))
1(d) l(t+1)=theta7+theta8*l(t)+theta9*ε(t+1)+theta10*(abs(ε(t+1))-sqrt(2/pi))

RM is the excess market return, ε(t)is a normal i.i.d. error term with zero expectation and unit variance, and mu(t) is the one-period expected excess return (conditional on short run(s) and long run volatility (l). log-volatility in equation (1b) is the sum of two components s(t) and l(t). Each component is an AR(1) processes with its own rate of mean reversion. l(t) is the slowly mean-reverting, long-run component and s(t) is the quickly mean-reverting, short-run component (theta4< theta8). They normalized the unconditional mean of s(t) to be zero.
So, the authors are estimating parameters theta1-10 from the following three equations (basically rearranging the ones above):

Rm(t+1)=theta1+theta2*s(t)+theta3*l(t)+sqrt(v(t))*ε(t+1),
s(t+1)=theta4*s(t)+theta5*ε(t+1)+theta6*(abs(ε(t+1))-sqrt(2/pi)),
l(t+1)=theta7+theta8*l(t)+theta9*ε(t+1)+theta10*(abs(ε(t+1))-sqrt(2/pi)),
where v(t)=exp(2(s+l))

My guess is this is some sort of component Arch model, it is just I have no clue how to code it in Eviews. If somebody could help me with this, I would be extremely grateful. The major problem is in order to replicate the rest of the paper, I need to get those short and long term volatility components.
Here is the link to the paper in case something is unclear from the description: http://www.newyorkfed.org/research/staf ... /sr254.pdf (page 4, 6 for equations and page 39 for table 1).
I hope somebody can help me with this (at least give me some hint). If you need further clarification, please do write to me. Thanks a ton in advance!

[biubiu]

Hi,
I am trying to do exactly the same thing -- replicating the component-garch model by Adrian Rosenberg 2008. I tried that in R by writing codes, but still got some problems. I saw you said the author suggested to solve it in Eviews. I was wondering if you had figured out how to do it in Eviews?