Bai and Perron str break test using Non-linear estimation

[startz]

Guys...think I am getting crazy..Thats what I found in a published paper:

"Nonlinear least-squares estimates of the following equation: i=(1-ρ)(β+γ*p(t)+δ*y(t))+ρi(t-1)
where i is the Federal Funds Rate, p is forecasted inflation, and y is the output gap. "

First of all, the above model is linear or non-linear???? if its non-linear how do we estimate it in Eviews??

Much appreciated

The model can be equally well fit as a linear model. The intercept will be (1-ρ)β, the coefficient on p will be (1-ρ)γ etc.

[lakis]

I definetly agree with you..to me this is a linear model...but with hell one should treat it as a nonlinear and estimate it like that ???

[startz]

It lets you get the long-run coefficients directly.

[trubador]

Unfortunately, there are certain cases that one should estimate nonlinear relationship directly instead of linearizing it. But, this is not one of them. Startz and Glenn have already pointed you in the right direction. So, all you have to do is get your hands dirty:

Code: Select all

wfcreate m 1960 2014 !rho = 0.8 !beta = 0.5 !gamma = 1 !delta = 2 series x1 = nrnd series x2 = nrnd series y = nrnd smpl @first+1 @last y = (1-!rho)*(!beta+!gamma*x1+!delta*x2)+!rho*y(-1) + nrnd coef(4) lcoef equation eq_linear.ls y = lcoef(1) + lcoef(2)*x1 + lcoef(3)*x2 + lcoef(4)*y(-1) coef(4) ncoef equation eq_nonlinear.ls y=(1-ncoef(1))*(ncoef(2)+ncoef(3)*x1+ncoef(4)*x2)+ncoef(1)*y(-1) coef(4) lcoef_e lcoef_e(1) = (1-ncoef(1))*ncoef(2) lcoef_e(2) = (1-ncoef(1))*ncoef(3) lcoef_e(3) = (1-ncoef(1))*ncoef(4) lcoef_e(4) = ncoef(1) matrix(4,2) compare colplace(compare,lcoef,1) colplace(compare,lcoef_e,2) show compare

[lakis]

Omg....You are absolutely right...I was just blind...thanks for the help all of you guys..

Much appreciated.xx

[EViews Glenn]

Note that if you want to get the long-run coefficients with standard errors after running the linear form, just put the expressions in the Wald test.