Dear colleagues:
I'm trying to repeat (using attached data in format Excel 4.0) the estimate roots obtained in the stochastic unit root model given by equations (expressed in state space form):
Dyt = b + f2Dyt-2 + dt[yt-1- b(t-1)-f2yt-3] + et et i.i.d. (0,s2)
dt = dt-1 + zt zt i.i.d. (0,w2)
The values of parameter estimated in the article of Leybourne, Mc Cabe y Mills (published in Journal of Forecasting, Vol. 15 Iss. Nº 3, pág. 267) are:
b = 6.1*10E-5 f2 = 0.148 w2 = 7.6*10E-7 s2 = 6.3*10E-5
Using previous estimates, and Kalman filter, they get, in each period, valued estimates about the root at = 1+ dt (the root has unit mean). The estimates always fluctuate between 0.94 and 0.96 while I get estimates oscillating about 1.
I think I make a mistake when the model is expressed in state space form and I am thanked for some help about this problem (also, I don't understand as, if the root has an unit mean, the estimates are always under the mean). I used the package SSF to estimate the roots in state space form (programmed by Sephard , Koopman and Doornik) implemented in Ox.
Best regards for your attention and thank's for your collaboration.
Román Mínguez
Dpto Métodos Cuantitativos para la Economía
Universidad San Pablo-CEU
MADRID
[log in to unmask] or [log in to unmask]
