Maximum likelihood estimation

[beniam]

I desperately need some assistance on the Likelihood Estimation Problem that I am encountering while etimating a 3 month interet rate uing an interest rate model.

I am using EVIEWS 7, Version 7.2.

The logl specification is as shown below.

'ckls Model

@logl logl1

@byeqn

'@param c(1) 0.002 c(2) -0.0273 c(3) 0.07 c(4) 1.35

error=r3m-exp(c(2))*(lagr3m)-(c(1)/c(2))*(exp(c(2))-1)

mttsq=(c(3)/(2*c(2)))*(exp(2*(c(2)))-1)*((lagr3m)^(2*(c(4))))

logl1=-(2*log(sqr(mttsq))+(error^2/mttsq))

While running for the estimation, it always returns to me an error message reading " Missing values in @LOGL series at current coefficients at observation 1". I have done a quick analysis on the data and the expressions in my LOGL object are totally out of the line. The "error", "mttsq" and "logl1" functions have no real values (i.e. it is missing and says NA). I do not know what to do and any help would be appreciated.

Thank you in advanvce!
Regard,
beniam

[EViews Gareth]

What starting values are you using?

[beniam]

What starting values are you using?

Gareth, Thank you so much for your replay. I am using known values from previous estimations (by other study) but I have even arbitrarily set the coefficients and none of it seems working. I appreciate your help very much!
Beniam

[EViews Gareth]

You need to find starting values for which the expression:
error=r3m-exp(c(2))*(lagr3m)-(c(1)/c(2))*(exp(c(2))-1)
does not result in an NA

[beniam]

You need to find starting values for which the expression:
error=r3m-exp(c(2))*(lagr3m)-(c(1)/c(2))*(exp(c(2))-1)
does not result in an NA

Thank you indeed for the help!

[beniam]

Hi,
I have succesfuly estimated two non-linear nested models using the Logl object and I wanted to apply the log likelihood ratio test. I have entered the following code in to the comand bar;

scalar lr = -2*(CKLS.@logl - Vasicek.@logl)
scalar q =1
scalar lr_pval = 1 - @cchisq(lr,q)

Where "Ckls" is the alternative model and the "Vasicek" is the null model.
While trying to estimate, I get an error mesage reading "LR is not defined". Am I doing something wrong? I appreciate any help.

Kindly,
Ben