I tested the calculated table and it works fine. However seems like my data isn't that clean and there are some changes in the salaries which are not raises. There 1 month increases such as bonuses and some 1 month decreases in some cases.
I like the calculated column solution although I didn't try it.
Here is more realistic look of my data. The last column called "Event" is something that I wrote manually.
In this case the formulas return the last change date as 01.10.2017, however a person can identify that the last raise was on 01.08.2017.
It seems like this case is a serious challenge. Do you have any ideas how to resolve it?
I'm sorry I didn't made it clear. The [Event] column is something I write manually just to demonstrate how changes could be something different than a raise. I don't have [Event] column in the original dataset.