Smooth Transition
Autoregressive Model
Eni Sumarminingsih
Smooth Transition Autoregressive
Model
• For some process, it may not seem
reasonable to assume that the threshold is sharp
• Smooth Transition Autoregressive (STAR) Model allow the
• Consider the special NLAR model given by
• If f() is a smooth continuous function, the autoregressive coefficient (α1 + β1) will change smoothly along with the value of Yt-1
• There are two particularly useful forms of the STAR model : the
Logistic STAR and the Exponential STAR
• The LSTAR Model generalizes the
standard AR model such that the AR coefficient is a logistic function :
• where
• is called the smoothness parameter • In the limit, as --> 0 or ∞, LSTAR
• For intermediate value of , the degree
of autoregressive decay depends on the
value of Yt-1
• As Yt-1 -, 0 so that the behavior of
Yt is given by
• As Yt-1 +, 1 so that the behavior of
Yt is given by
• Thus the intercept and the AR coefficient
smoothly change between these two
extremes as the value of Yt-1 changes.
• The ESTAR model uses
, > 0
• As approach zero or infinity, the model becomes an AR(p) model since is constant
• Otherwise, the model display nonlinear behavior
• As Yt-1 moves further from c,
approach 1 behavior of Yt is given by
Eni Sumarminingsih, SSi, MM
Test for STAR Model
Step 1 : Estimate the linear portion of the AR(p) model to determine the
order and to obtain the residual {et}
Test the significance of the entire
regression by comparing TR2 to the critical value of 2.
If the calculated value of TR2 exceed the critical value from a 2 table,
reject the null hypothesis of linearity and accept the alternative
hypothesis of a smooth transition model.
Alternatively, you can perform an F
Step 3 : If you accept the alternative hypothesis (i.e., if the model is
Uji F
• Hipotesis nol: restricted model valid
• Menduga restricted model dan unrestricted model
• Memperoleh JK Galat untuk restricted model dan JK Galat
untuk unrestricted model, dan menghitung statistik uji F.
JKGR: JK galat restricted model JKGU: JK galat unrestricted model
kU: jumlah peubah eksogen (termasuk konstanta) pada unrestricted model