I would like to perform a difference-in-differences regression to determine the effect of the national living wage (NLW) on income inequality. I would like to use the log50/10 wage ratio as my measure of inequality.

There are 2 time periods I am interested in: before and after the introduction of the NLW. I would like to see how the log50/10 ratio changed as a result of the NLW. But the log50/10 ratio is constant within each time period. Without variation, how am I supposed to run this regression?

I have panel data which records hourly wage pay for 4900 individuals, and a time dummy for before/after NLW and a treatment dummy for treatment/control as well as an interaction term. How can I measure to how the hourly wage pay has increased since the NLW but for different percentiles of hourly pay. I am expecting the 10th percentile individuals to have the largest increase in wages and want to see how far the spillover effect is up to the 50th percentile.

Do I need to run 5 separate regressions? For each decile? And then compare them? How can I even code this on Stata?

Thanks