I am applying Bai and perron (1998) structural change model in eviews using sequential l+1vs l. It's giving me different break dates in sequential and repartition method. Which one to consider.
The Bai-Perron test is a multiple structural breaks test. Using the sequential approach, the significance of an additional break (whose date is a priori unknown) is tested at each step of the procedure. The total number of breaks to consider is the one for which the F-statistic is significant. So there will be at least two break dates to consider, since this is a multiple break test.
However, I do not recommend using the sequential approach. Instead, use the parsimony criterion minimization approach, as Yao (1988) has shown that the latter approach provides a more consistent estimate of the true number of breaks in the series. I can recommend one of my papers where I used the LWZ parsimony criterion (Liu et al., 1997) which is a modified version of the Schwarz criterion; the difference between these two criteria is just in the severity of their penalty.
You can find the paper on my publications portal. It will also help you to orientate the interpretation of the test results.