How to model a quadratic trend + independent dynamics #29
cdriveraus
started this conversation in
General
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Was emailed this Q, seems of general interest so here's a demo showing how to use the nonlinear specification capacities of ctsem to estimate a quadratic trend with a separate dynamic process occurring 'around the trend' as such. Can alternatively think of as a dynamic residual. In the last section I show an exponential trend, which does not require the PARS specification or recompilation, but will only work for trends that do not show any U shape.
Model (via ctModelLatex() ) and plot for the quadratic + dynamics model
Beta Was this translation helpful? Give feedback.
All reactions