PSTR model

Hello,

Please I would like to apply panel smooth transition autoregression model (PSTR) model, which is developed by Gonzalez et.al.,(2005). http://swopec.hhs.se/hastef/papers/hastef0604.pdf

So, I was wondering is it possible to apply it using GAUSS ? if it is possible, would you mind if you provide me with the codes for this model ?

Thanks in Advance.

Regards,

Nermeen

1 Answer



0



Gonzalez, et al., solve the problem using an iterative nonlinear least squares method. The problem is complicated by constraints on some of the parameters, in particular with a logistic specification requiring gamma > 0, and c1 <= c2 <= ... <= cm.

I would recommend solving the problem as a maximum likelihood problem because the statistical inference is straightforward. Using Constrained Maximum Likelihood MT (CMLMT) an Application available from Aptech Systems, you would be able to enforce the constraints on the parameters, and the covariance matrix of the parameters is returned as a byproduct of the estimation. You would need to provide a procedure for calculating the log-likelihood, and another procedure computing the inequality constraints.

Your Answer

1 Answer

0

Gonzalez, et al., solve the problem using an iterative nonlinear least squares method. The problem is complicated by constraints on some of the parameters, in particular with a logistic specification requiring gamma > 0, and c1 <= c2 <= ... <= cm.

I would recommend solving the problem as a maximum likelihood problem because the statistical inference is straightforward. Using Constrained Maximum Likelihood MT (CMLMT) an Application available from Aptech Systems, you would be able to enforce the constraints on the parameters, and the covariance matrix of the parameters is returned as a byproduct of the estimation. You would need to provide a procedure for calculating the log-likelihood, and another procedure computing the inequality constraints.


You must login to post answers.

Have a Specific Question?

Get a real answer from a real person

Need Support?

Get help from our friendly experts.