Test if the regression results differ significantly between two periods

[Ama Perera]

I am working with a panel dataset comprising 40 countries, covering quarterly data from 1990 to 2020. In this dataset, I have two main variables: 'risk' (independent variable) and 'policy' (dependent variable). Both variables are observed on a quarterly basis.

My objective is to investigate whether there are variations in the baseline regression results between the 'risk' and 'policy' variables based on changes in political movements within individual countries. To capture political movements, I've included a variable named 'polity,' which is measured annually.

For instance, let's consider the country 'Belgium.' In 2014, there was a notable political movement that resulted in a change in the 'polity' value assigned to Belgium, specifically from 2013 to 2014.

To assess the impact of this political movement, I aim to examine whether the regression results for 'risk' and 'policy' differ significantly between the pre-movement year (2013) and the post-movement year (2014).

One approach I am considering is running separate baseline regressions for Belgium in 2013 and 2014. Subsequently, I want to test whether the coefficients from these two regressions are statistically different.

Could you please suggest an approach and provide relevant Stata code to conduct this analysis?

Thank you.

[Ama Perera]

or would a difference in difference approach be better?

For example.


gen post_polity= (year == 2014) & (country == "Belgium") *
gen interaction_term = risk* post_polity
reg policy risk post_polity interaction_term if country == "Belgium" & (year == 2013 | year == 2014), robust