Parallel Trend Assumption Graph

[Jheelum Sarkar]

Hi everyone,
I am conducting difference-in-difference analysis to find impact of a natural event in the year of 2013 on household level data. I am using line graph to check the parallel trend assumption with average monthly per capita consumption expenditure (MPCE) as main variable of interest. The graph looks like the following:

Can anyone please explain me if it is okay for the two lines to be parallel after 2013? Here the blue line is for treatment and red line is for control group. Any advice/comment will be helpful

[Jared Greathouse]

Yeah it's okay, it just means there's not much of a treatment effect.


My only question for you, is is it possible for you to get more pre-intervention years of data?

[Jheelum Sarkar]

Thank you so much for the kind reply!!!! I can get another pre-intervention year of data but that correspond to 2004-05 period, 5 years prior to 2009-10. Is it okay to take this?

[Jared Greathouse]

You're better off not having it then, in my opinion.

[Jheelum Sarkar]

Thank you for the kind reply. May I know if it is okay to proceed with this existing graph?

[Jheelum Sarkar]

I made some adjustment in the treatment and control group districts and got the following graph:

Please advice me if it is okay to proceed with this?
Also, I conducted the Difference-in-difference analysis too (here MPCE_new is the monthly per capita consumption expenditure, treat_uk_new is the treatment group, time is the time dummy and did_uk_new is the DID variable (treat_uk_new*time). The result is in the following:

Please advice me on this. Thank you

[Jared Greathouse]

Yep this looks fine, the only thing I'd change is

reg MPC i.treat_uk_new##c.time, r

[Jheelum Sarkar]

Thank you so much for the advice and the suggestion!