Page 1 of 1

Bugs in BREAKLS

Posted: Mon Dec 02, 2013 6:38 am
by svilen
We found two bugs in BREAKLS.

1. It doesn't use the user specified breaks, but uses dates which are two years after the specified breaks.

Code: Select all

%f = "test_file" close {%f} wfcreate(wf={%f}) q 2000 2015 smpl 2001 2015 series s1 = nrnd series s2 = nrnd series s3 = nrnd smpl 2002 @last equation eq1.BREAKLS(method=user, breaks="2006q1 2011q4") s1 s2 s3 show eq1
2. With the code above try to change the method to sequential tests of single l+1 versus l breaks. If you use the GUI EViews crashes. But if you do it with code it doesn't crash:

Code: Select all

equation eq1.BREAKLS(method=seqplus1) s1 s2 s3
ps it seems that there is a typo on page 283 of EViews 8 Command and Programming Reference
-> breakls(method=user, breaks=”1990q1 2010q4”) m1 c @nv unemp

Re: Bugs in BREAKLS

Posted: Mon Dec 02, 2013 10:33 am
by EViews Glenn
Thanks for the detailed description. We'll take a look.

[edit] - Fixed the crash and a related bug in the parsing of user-specified dates in some equations estimated with sub-samples. The fixes will appear in the next update. Sorry for the inconvenience.