Chapter 13; page 39; paragraph 3 "...Where we have an incurred claims triangle missing all data prior to a certain calendar year we can derive development factors in a similar way to a full triangle using the column sum average" By column sum average doesn't this just mean the development factor (summing all cohorts) at that development period? So if the triangle looks like the one below, for period 2->3 we could calculate this as (10+30) / (5 + 20) ----- 1--2--3--4--5 2001 # # 15 17 17 2002 # 5 10 14 2003 10 20 30 2004 12 14 2005 11
That's correct Cartland. You only include items in the column sum average if they have an entry in both columns. So you correctly excluded 15 from the numerator as there was nothing to put in the denominator.
