Re: [Gretl-users] interpreting the “StrucBreak” package for gretl

[Sven Schreiber]

Am 15.10.2018 um 11:31 schrieb Yusuf Abduwahab Hassan:
>
> Can i safely conclude that there is no evidence of structural break in 
> the chosen dates?

No I don't think so.

> ? SB_Tests(&bII)
>
> ===================================================================
>
> OUTPUT FROM THE TESTING PROCEDURES
>
> ===================================================================
>
> a) supF[Y.A.H1] <#_msocom_1>tests against a fixed number of breaks
>
> --------------------------------------------------------
>
> supF(1|0)supF(2|0)supF(3|0)
>
> 11.43211.18910.897
>
> Critical values:
>
> supF(1|0)supF(2|0)supF(3|0)
>
> 10%10.379.438.48
>
> 5%12.2510.589.29
>
> 2.5%13.8611.6310.14
>
> 1%16.1912.9011.12
>

Testing none against 2 or 3 breaks is significant here at "conventional" 
levels. Against 1 break it is borderline (10%, but not 5%).

> --------------------------------------------------------
>
> b) Dmax tests against an unknown number of breaks
>
> --------------------------------------------------------
>
> UDmax test: 11.431741
>
> Crit. values:10%: 10.86 5%: 12.59 2.5%: 14.15 1%: 16.19
>
> ........................................................
>
> WDmax test(crit. val.)
>
> 10%13.3311.71
>
> 5%14.3713.66
>
> 2.5%14.9015.33
>
> 1%15.8717.80
>
> ********************************************************
>

Both UDmax and WDmax again borderline, and as it says, against an 
unspecified (unknown) number of breaks under the alternative hypothesis.

> supF(l+1|l) tests using global optimizers under the null
>
> --------------------------------------------------------
>
> supF(2|1) 10.931981
>
> supF(3|2) 10.701993
>
> --------------------------------------------------------
>
> Critical values:10%5%2.5%1%
>
> supF(2|1)12.19 13.83 15.51 17.58
>
> supF(3|2)13.20 14.73 16.55 18.31
>

These test results are not significant.

So taking these results together the evidence is mixed in my view.

cheers,
sven

Am 15.10.2018 um 11:31 schrieb Yusuf Abduwahab Hassan: Can i safely conclude that there is no evidence of structural break in the chosen dates? No I don't think so. ? SB_Tests(&bII)   ===================================================================    OUTPUT FROM THE TESTING PROCEDURES =================================================================== a) supF[Y.A.H1]  tests against a fixed number of breaks --------------------------------------------------------     supF(1|0)   supF(2|0)   supF(3|0)       11.432      11.189      10.897   Critical values:            supF(1|0)   supF(2|0)   supF(3|0) 10%           10.37        9.43        8.48 5%            12.25       10.58        9.29 2.5%          13.86       11.63       10.14 1%            16.19       12.90       11.12 Testing none against 2 or 3 breaks is significant here at "conventional" levels. Against 1 break it is borderline (10%, but not 5%). -------------------------------------------------------- b) Dmax tests against an unknown number of breaks -------------------------------------------------------- UDmax test: 11.431741 Crit. values: 10%: 10.86 5%: 12.59 2.5%: 14.15 1%: 16.19   ........................................................        WDmax test    (crit. val.) 10%       13.33         11.71 5%        14.37         13.66 2.5%      14.90         15.33 1%        15.87         17.80 ******************************************************** Both UDmax and WDmax again borderline, and as it says, against an unspecified (unknown) number of breaks under the alternative hypothesis. supF(l+1|l) tests using global optimizers under the null -------------------------------------------------------- supF(2|1)     10.93  1981 supF(3|2)     10.70  1993 -------------------------------------------------------- Critical values:   10%    5%  2.5%    1%        supF(2|1)  12.19 13.83 15.51 17.58        supF(3|2)  13.20 14.73 16.55 18.31 These test results are not significant. So taking these results together the evidence is mixed in my view. cheers, sven