Hi friends,

I need some advise. I am trying to estimate the effect of government program (funding) on regions at metropolitan statistical area (MSA) level. The funding is very selective (i.e., so far only 1/3 of MSAs got this funding) and lasts generally 5 years. As the funding lasts about 5 years, each year has a cohort of about 10 new entrants. Each recipient gets different amount of funding at different period. This funding ($) is the key independent variable. The dependent variable is continuous variable ($). I constructed a panel data with several control variables between 2006 and 2018.

These days, I suddenly remembered that I read an article (Png 2017) saying...
"To investigate the impact of the UTSA on R&D expenditure, I apply an empirical strategy of difference in differences (Bertrand, Duflo, & Mullainathan, 2004), with multiple treatments at different times with different intensity in the various states. Specifically, I estimate the following model, for company i, in state s, in year t:

ln (1+ Rist) = b1*UTSAst + b2*Xit + bis + bt +eist (1).

Rist represents R&D expenditure by company i in state s in year t, UTSAst represents the increase in the legal protection of trade secrets arising from the UTSA being in effect, Xit are time-varying company characteristics, bis, bt are company by state and year fixed effects, and eist is an idiosyncratic error term."
I think my panel is similar to the case above. But I have no idea how to implement this. I will appreciate any comment. Thank you.