Panel data model - Interaction term

#1
I do a panel data regression on stock returns to examine the effect of ESG during a crisis and during normal times. My dataset is dated from Jan 2021 back to Jan 2017. My dependent variable is stock return, and the independent variable of interest is an ESG-variable (values from 0-100).

My main regression equation:
Return = ESG + Fama-French Factor model + some firm speficis charackters + time fixed effects + industry fixed effects.
In this regression, ESG is insignificant in explaining the returns.

In my second model, I add dummy interaction terms for two specific periods to examine if the effect of the ESG-score is significant in specific periods compared to the overall period. Now, my model is as follow:
Return = ESG + ESG*Crisis + ESG*Post-crisis + Fama-French Factors + some firm specific charackters + time dummies + firm fixed effects.

Crisis period: February- March 2020. Post-crisis period: April 2020 - Jan 2021.
In this regression, the ESG-variable is still insignificant. ESG*Crisis is significant with a positive coefficient, while ESG*post-crisis is significant with a negative coefficient.

How should I interpret these results?
Does it make sense to interpret the interaction terms when ESG itself is insignificant?
 
Last edited: