time-varying kalman filter

[ege_man]

Dear Tom,
I looked at the example files of Baltagi's Econometrics Textbook, the examples given are at the introductory level based on OLS estimations. dlny/dlnx is the elasiticity that I can get if I estimate the model in levels. But if I use log first difference the parameters give me change in elasticity difference of dlny/dlnx therefore we need to take integration to retrieve elasticities. Is there any way to do this with RATS.
The code I used already based on local level model am I right? Therefore we are already estimating the model with local linear trend component. In that case do results show us detrended results?
Thanks

[TomDoan]

Dear Tom,
I looked at the example files of Baltagi's Econometrics Textbook, the examples given are at the introductory level based on OLS estimations. dlny/dlnx is the elasiticity that I can get if I estimate the model in levels. But if I use log first difference the parameters give me change in elasticity difference of dlny/dlnx therefore we need to take integration to retrieve elasticities. Is there any way to do this with RATS.

First of all, I think you're missing the point about the Baltagi reference. Note how effectively everything in it is rather carefully adjusted to eliminate/reduce trends that don't respond to price signals. You have to do that in order to (properly) estimate elasticities from time series data. One should not make disparaging comments about least squares. Often complicated models are used to obscure otherwise really bad empirical work.

Also, you're wrong about the elasticities. If you do symmetrical filters on the left and right, you're still estimating the elasticity.

The code I used already based on local level model am I right? Therefore we are already estimating the model with local linear trend component. In that case do results show us detrended results?

No. TVC has nothing to do with local level/local trend (unless you're only doing a "regression" on an intercept).

[ege_man]

Dear Tom,
Thanks. I read the example in the textbook again and then applied ADF test to the each series obtained from filter(remove=trend) procedure. The results suggest that all variables are difference stationary, therefore first-differenced model is appropriate.
Regards

[ABDHUT123]

Dear Tom
is there any way to estimate the time-varying parameter model with endogenous regressors in RATS as done by Kim and Nelson 2006 in their paper "Time-varying parameter models with endogenous regressors".