Zivot Andrews breakdate

[カルナニディ，クリスクマール]

Hi there,

Currently running the Zivot Andrews unit root test and comparing it with the results from eViews, Stata and R (ur.za)

Across all four tests, the t-statistics are the same, however break dates are different.
Assuming the break date for a particular variable is 2010 in eViews and Stata, the break date for R(ur.za) and Python(Statsmodels) would usually be 2009, which is 1 year before whatever date that the other program gives.

Is this normal?

Thank you in advance!

Best Regards,
Kris