EViews.com

I would like to exploit the "compromise" benefits of PMG estimation and estimate the ARDL Co-I relationship between real GVA levels for each of 12 industrial sectors and national GDP/GVA. Given the common long-run coefficient estimates in the PMG method (ie a unit coeff on GDP to ensure sectors add-up) is there any smart way to also allow for long-run trends in sector GDP shares (eg Manuf share -ve, Services share +ve). I presume the PMG method itself may be complicated by a (common) deterministic trend in the long run relationship ? Any guidance on how to model long-run sector share trends with PMG appreciated.

Would this .PRG code achieve what is desired ?
@trend*@expand(country)

From this post below:

Re: Adding a trend for each cross-section in panel analysis?
Report this post
Quote
Postby DCFerral » Wed Apr 10, 2019 8:05 am
Try @trend*@expand(country)

That will create a separate trend coefficient for each cross section