Dear Statalist,
I'm currently analyzing a price database and I wanted to try to use structural breaks tests.
Attached file contains a sample series.
I run this simple (and very raw) regression looking for trend and seasonal effects with 12 dummies (d1-d12)
My idea was that 150 observation could be enough to run this test, but STATA has a different opinion.
Can someone please explain me why?
I am aware that running a different code, let's say:
I would obtain a result instead.
Test for a structural break: Unknown break date
Number of obs = 150
Full sample: 2003m1 - 2015m6
Trimmed sample: 2004m12 - 2013m8
Estimated break date: 2009m7
Ho: No structural break
Test Statistic p-value
-----------------------------------------------
swald 110.2370 0.0000
-----------------------------------------------
Exogenous variables: t
Coefficients included in test: t
What's my best chance to incorporate this result in another regression?
Many thanks,
Valerio
I'm currently analyzing a price database and I wanted to try to use structural breaks tests.
Attached file contains a sample series.
I run this simple (and very raw) regression looking for trend and seasonal effects with 12 dummies (d1-d12)
Code:
regress mp_price2 t d1-d12, noconstant estat sbsingle, breakvars(t)
Can someone please explain me why?
I am aware that running a different code, let's say:
Code:
regress mp_price2 t, noconstant estat sbsingle, breakvars(t)
Test for a structural break: Unknown break date
Number of obs = 150
Full sample: 2003m1 - 2015m6
Trimmed sample: 2004m12 - 2013m8
Estimated break date: 2009m7
Ho: No structural break
Test Statistic p-value
-----------------------------------------------
swald 110.2370 0.0000
-----------------------------------------------
Exogenous variables: t
Coefficients included in test: t
What's my best chance to incorporate this result in another regression?
Many thanks,
Valerio