What steps should I take to correctly code for estimates that look at the effects of floods on unemployment rate? I’m using TWFE Difference-in-difference with staggered treatment adoption method. I’m examining all 54 countries in Africa and looking at all floods and droughts each year between 2005 and 2020. I have an excel file with all the countries, the start year (when the flood happened and which country), the treatment/dummy variable flood occurrence (1 if flooding happened in that country in a specific year, e.g: flooding in morrocco in 2020), the unemployment rate for each country each year, and a bunch of covariates.

I designated the treated countries as those above the 90th percentile and untreated countries as those below the 90th percentile. I found parallel trends in the average unemployment rate at the national level for countries above and below the 90th percentile. However, not sure if I’m thinking correctly here or not..