Run regression separately for several periods

[Helen Chang]

What are some reasons that can justify the necessity to run regressions after dividing the data into several periods?
e.g., instead of run reg xxxxx for the whole sample period, I will run reg xxxxxx for the first half 10 years and run the regression again for the latter half separately.

[Fei Wang]

Helen, you may do so if you think the association between the dependent variable and the independent variables changes from the first half period to the second half.

[Carlo Lazzaro]

Helen:
why not running a regression with one more categorical predictor (0=years_1_5; 1=years_6_10)?