ask for help

[maoqi]

Does anyone know how to input the function into eviews? Please give me some suggestions. It would be better to tell the whole functions for "system" in eviews. Thank you very much

[maoqi]

My function is in the attachment. Thanks

[EViews Glenn]

Do you only have two errors or is that a mistake?

[maoqi]

yes, only two erroes. The 2 errors are for slt-1 and skt-1. And I think errors are not input into EViews. Thank you please tell me how to input the function into EViews. I'd appreciate it very much.

[EViews Glenn]

You'll create a system object and specify each equation explicitly in terms of the coefficients and variables. For example, the first equation might look like

Code: Select all

d(sl) = c(1)*d(sl(-1)) + c(2)*d(sk(-1)) + c(3)*d(w) + c(4)*d(q) + c(5)*d(ph) + c(6)*d(y) - c(7)*sl(-1) - c(8)*sk(-1) - c(9) - c(10)*w - c(11)*q - c(12)*ph - c(13)*T

I've done these all using the C coefficient vector, but you might like to define your own, say a Beta, an Alpha, and a Pi. Since there are no restrictions on your coefficients, you'll use a different one for each variable and equation.

[maoqi]

Thank you very much. But how to define coefficients Beta and Alpha?

[EViews Glenn]

Code: Select all

coef(10) beta

[maoqi]

Thank you very much. I am now able to test the function. However, I am not sure about my results and I still have some difficulties. Would you like to give me some advices and tell me how to reach the partial elasticities? The results and the partial elasticity question are in the attachment. Thank you very much.

[EViews Glenn]

What estimation method did you choose? You can't tell from the output, but it looks to be iterative. I suspect that's not what you want.

[maoqi]

the method to calculate from the coefficients to the partial elasticities is in the last paragraph on page 122 to the top of 123. It is a paper of Holly's. The relevant PDF document of the paper is attached with all the approaches to the model and restrictions. I just use the different data.
Thank you very much. I really have difficulty. I'd appreciate it very much.

[EViews Glenn]

Your request to look at the paper are outline calculations is far beyond the normal technical support that we provide so regretfully, I'm going to have to stop here unless you have more specific questions about how to do a particular calculation in EViews. I hope that you understand that we can't spend all of our time answering questions like this, otherwise there would be no future versions of EViews.

My last comment was about your estimation results was simply to point out that they didn't converge and that I wasn't sure why convergence would be an issue since nothing in your estimation indicated that you should need iterative estimation.

[maoqi]

I choosed full information maximum likelihood. I don't think the results need convergence. Should the resutls be convergent?

[EViews Glenn]

They do if you are estimating with FIML. Least squares won't require convergence. Or at the very least, estimate with least squares to get starting values, then estimate with FIML.

[maoqi]

Thank you very much. And is there a command in EViews to calculate the elasticities in the attachment? I am not familiar with EViews. Please do give me a clue. I'd appreciate you very much.

[trubador]

Actually, the method used in the study is a Vector Error Correction model with imposing restrictions, which can easily be estimated in EViews. However, EViews does not allow exogenous variables in the cointegrating equation (i.e. eq 12 in the paper). I cannot think of any easy or simple way to estimate this model...