News Impact Curve

[Sam2010]

Hello,
I'm estimating an EGARCH (1 1) model and the News impact curve using the attached program codes, but all 6 News impact curves seem to be the same. Can anybody tell me what I'm doing wrong. See that attached program codes. I will appreciate any help I can get. Thank you

Using Eviews 7.0 but updated to 7.2 recently.

Sam

[trubador]

First of all, there is no way to detect your errors (if any) without seeing "your" code, since the document you attached contains the original version of the code. It is an example code written only for demonstration purposes and further modification is needed for a more customized use. If you have your own estimated EGARCH model, then you should skip the simulation part and use only the last part of code that generates the NIC. However, before doing that, you should compute and save the necessary inputs (residuals, conditional variance, standardized errors, estimated coefficients, etc.).

[Sam2010]

I'm using EGARCH from the window-quick-estimation-Arch-EGARCH and not command window. I have identified residuals, conditional variance, standardized errors, estimated coefficients in the window and my problem now is running the News impact curve potion of the program from command window while the data are in the work file window. Thank you

[akram hasanov]

Dear Trubador,

I have written following code for news impact curves of GARCH and EGARCH following the paper Engle and Ng (1993). Would you please have a look once you are available whether it is OK.

Thanks
Akram
Universiti Putra Malaysia

The code is as follows

'change path to program path
%path = @runpath
cd %path
' load workfile
load aaaaa.wf1

' set sample
smpl 1/06/2003 12/09/2010

'find the value of A
scalar a = garch11.C(2)+garch11.c(4)*(@stdev(returns))^(2)

'generate ht series for news impact curve
series ht_news_garch = a+garch11.C(3)*returns(-1)^(2)

'find the value of A1
scalar A1=(@stdev(returns)^(2*egarch.c(5)))*exp(egarch.C(2)-egarch.C(3)*(2/(22/7))^(0.5))

'assign lagged return series
'series lagged_return_egarch = returns(-1)

'generate ht series for news impact curve
series ht_news_egarch = @recode (returns(-1)>0, a1*exp(((egarch.C(4)+egarch.C(3)) / @stdev(returns) )* returns(-1)), a1*exp(((egarch.C(4)-egarch.C(3))/@stdev(returns))*returns(-1)))

group news_impact returns(-1) ht_news_garch ht_news_egarch
news_impact.scat