Hi,
I am comparing Fama-French 3-Factor model and Fama-Carhart 4 on 40 industry returns. I am trying to do Sub-sample analysis with the monthly data from 1963-2018 based on Business cycles identified by the CFNAI index.

However, my main problem is that each recession periods in history has different lengths. Some recession period lasts more than 20 months, some only has 12 months. Therefore, if I do a sub-sample analysis on the recession periods, each sub-sample has different lengths; would it affect my statistical inferences and the power of my regression results at the end? Since some data only has a few observations.

Is there another method where I can account for the business cycle effects on Asset Pricing models?

Thanks in advance.