if claims inflation is 2% per year, I don't need to inflate the cost to the mid point of when I think claims will be paid? Thanks!
For Accident Year data to policy year premium rating, we adjust to mid-point afwmter making uniform pattern assumptions. But, in this question data is sorted policy year and rate is also for a policy year. So, with 'all other factors being constant' assumption - you can make it by differencing policy year in years.
I will try to write in different way:- If you expect particular claims payment delay, say, 'x' from policy inception, and assumed patterns are similar. Policy year after y years will also have claims payment delay from policy inception 'x'. So, basically inflation needed for y years. No need for mid-point, this is applicable to any earning, incurring, paying pattern as long as remains same. Difference between claims payment date for policy year n and policy year m(where m>n) is m-n, when the pattern of earning, incurring and paying is same for all years. That is why 1.02^(5-1) is used to make 1st policy year value to as-at 5th and so for other years.