Validity of panle threshold model

[Devashish Singh]

i am using xthreg command to run a panel threshold model. my p value for single threshold is around 0.15, whereas p value for double threshold is around 0.05. can i apply the double threshold? any insights would be appreciated

[George Ford]

do the estimated double thresholds make any sense? is there some reason to expect three regimes? I suspect if there's high non-linearity, more thresholds would fit better.

[Devashish Singh]

yes theoretically double thresholds make sense. i tested for three regimes and the third one is insignificant. i am just confused if i can apply the double threshold given that the single threshold is insignificant. thank you for your assistance

[George Ford]

If theoretically you expect it, then I think yes. And if thing is really non-linear, then a single threshold may be miss it. You'll need to make a strong case for double.

The single isn't too far off in any case.

[Devashish Singh]

okay. thanks a lot for your help )