Regression with bai-perron break

For technical questions regarding estimation of single equations, systems, VARs, Factor analysis and State Space Models in EViews. General econometric questions and advice should go in the Econometric Discussions forum.

Moderators: EViews Gareth, EViews Moderator

justin_m
Posts: 13
Joined: Sun Feb 01, 2015 8:37 pm

Regression with bai-perron break

Postby justin_m » Fri Jun 02, 2017 12:05 am

When I run regressions with Bai-Perron global breaks, I get, for instance, two coefficients for a variable if there is one break. The coefficient in the second period is the value on top of the coefficient in the 1st period (as it would be if we just multiplied the variable by a dummy) or is it the total value for that period?

In many cases I have a negative coefficient in the 1st period and a positive coefficient in the 2nd period of similar magnitude, both significant. This could mean that the total effect in the 2nd period is zero or that indeed it is positive.

E.g. y= a+b1*x1+b2*x1*d+e, d = 1 if we are in 2nd period or d=0 if we are in the 1st period.

my question is whether the coefficient that eviews reports for the 2nd period is b2+b1 (the total effect for the 2nd effect) or simply b2 (the effect on top of b1).

Thanks.

justin_m
Posts: 13
Joined: Sun Feb 01, 2015 8:37 pm

Re: Regression with bai-perron break

Postby justin_m » Fri Jun 02, 2017 12:17 am

I see that the estimation command is:

Y= @BEFORE("8/11/2008")*C(1)*X1(-1) + @AFTER("8/11/2008")*C(2)*X1(-1) + C(3) + C(4)*X2(-1) + C(5)*X3(-1) + C(6)*X4(-1) + C(7)*X5(-1) + C(8)*X6(-1) + C(9)*Y(-1)


This suggests that the b2 is the effect in the second period rather than the effect on top of b1. That would be the case if:

y= C(1)*x1(-1) + @AFTER("8/11/2008")*C(2)*x1(-1) + C(3) + C(4)*x2(-1) + C(5)*x3(-1) + C(6)*x4(-1) + C(7)*x5(-1) + C(8)*x6(-1) + C(9)*y(-1)


correct?

EViews Glenn
EViews Developer
Posts: 2671
Joined: Wed Oct 15, 2008 9:17 am

Re: Regression with bai-perron break

Postby EViews Glenn » Fri Jun 02, 2017 5:37 pm

The representations view accurately reflects the parameterization.


Return to “Estimation”

Who is online

Users browsing this forum: No registered users and 32 guests