Hi Ekkehart

thanks for producing such a useful function. I have a query. When I specify a simple trend function of the type ln(y)=a +b*ln(Time)

the package produces a graph of a variable constant (a) coefficient and a constant trend value of (b).

Is it not possible to specify which coefficent should be variable so that the constant is a constant and the trend has a variable coefficient -which could be interesting as trend rate change may be important.

kind regards

Brian Revell

Professor Emeritus

Harper Adams University

On Mon, 2 Mar 2026 at 11:12, Ekkehart Schlicht <schlicht@lmu.de> wrote:

Hi all,

a new version of the TVC package doing time-varying coefficients
estimation in available on the Gretl server, along with an updated
version of the associated graphical backend TVCplots.

As a new feature, a Monto-Carlo check for testing time-variability of
regression coefficients has been added. This may also serve in some
cases to test the assumption of time-invariance in conventional models.
The feature is experimental. Comments are welcome.

Cheers

Ekkehart

Ekkehart Schlicht
Professor emeritus of Economics
Department of Economics
Ludwig-Maximilians-Universität München
www.semverteilung.vwl.uni-muenchen.de/ekkehart
ekkehart.schlicht@econ.lmu.de
Tel. +49 (0)8152 9149270
_______________________________________________
Gretl-users mailing list -- gretl-users@gretlml.univpm.it
To unsubscribe send an email to gretl-users-leave@gretlml.univpm.it
Website: https://gretlml.univpm.it/postorius/lists/gretl-users.gretlml.univpm.it/