kapetanios, shin & snell unit root test.

[mehul_98]

hi,

just wondering if theres any way i can carry out the KSS unit root test on eviews or if theres any other programme any one knows of that is capable of carrying out the test?

Thanks! :)

[Atish Kumar Dash]

Hi

I am also searching the procedure to run KSS unit root test in Eviews. If anyone knows, kindly let me know.

best regards
Atish

[adwy]

Hi.Is any one know how to use EVIEWS to run the KSS nonlinear Unit Roots test? Hope that anyone who know this technique and method can sharing with me.
Thank you.
Have a great day.

[neverhood]

Kapetanios, Shin & Snell themselves take a taylor approximation on their non-linear specification which yields an equation that you can easily estimate in Eviews:

Dy(t) = a*y(t)^3 + b*y(t-1) + c*y(t-2) + .... + error(t)

that's it, so you regress the difference to its cube and lagged terms. How do you determine how many lags to use? Start with 10-12 than reduce 1 by one and check Akaike Info Crit, pick the one yielding the smallest AIC.

Than of course, you just compare the statistic that you obtain for "a" with the table provided in KSS(2003)

[adwy]

I got it now. Thank you so much to you "Neverhood", you are so nice.

[C_lon]

Hi colleagues, I am in the process of performing the KSS test. However, I am not sure how to carry out hypothesis test in Eviews since the hypotheses are H1: a=0 vs. Ha: a<0, with a being the coefficient on the cube term? Can I use the Wald coefficent restriction option? If yes, how do I specify that the alternative hypothesis is a one-sided test?

[daniel_pl]

First, the above procedure is right. Second, if your number of observations is higher or less than 1000 than you need new critical values. Monte Carlo would have to take place.