EViews.com

Good evening

I would like to ask whether it is possible to derive impulse response functions from a threshold model. Thank you beforehand.

EViews does not currently compute impulse response functions for threshold models.

Thank you very much for the information.

Hey everyone, I had a quick question regarding conducting threshold analysis on categorical variables and would really appreciate any help.

As an overview, below are my data pieces and their respective types:

• Regular dependent variable (continuous data)
• The first independent variable is regular (continuous)
• I have 3 other independent variables, which are all categorical dummy variables.
o First dummy variable is gender, so three options (male, female, other)
o Second is highest education level, so around 4 options (the format in the model is =1, =2 =3 =4 etc)
o Third is family income group, which is 7 options

Now, if I don’t include the three categorical dummy variables, and simply put in the equation specification box “y c x,” then I can specify the threshold variable “x” and simply run a threshold regression.

However, once I start to include the categorical dummy variables using the “y c @EXPAND(variable1, @dropfirst) @EXPAND(variable2, @dropfirst) @EXPAND(variable3, @dropfirst)” in the equation specification, things don’t work.

My equation specification is:

"irv c age @expand(d1, @dropfirst) @expand(d3, @dropfirst) @expand(d4, @dropfirst)." If i then add in "age" in the threshold variable specification, and then If I click “OK” then it brings up an error saying "D3=1 is not in the equation for the list of regressor variables"?



My two questions are:

• Why does this occur? I can run a regular OLS regression using the exact same equation specification, and a break-point test.

• How can I run a threshold regression on the variable AGE (continuous data) while including the 3 categorical variables (no thresholds on these). The above equation works in OLS, just not with threshold?

Thank you!