Expense Trend vs Premium Trend

[bermi]

pg 136 Werner shows an Expense trend Diagram. On pg 137 it says - "Since the experience period is a calendar year, the average date the policies are written and earned is the same."

Pg 84 shows a CY EP historical period. The average written date differed from the average earned date - my understanding is because this is due to the historical data being "EARNED PREMIUM" and earned premium comes from policies:
- written in the historical period
- written prior to the historical period (based on policy duration)

Please relate pg 137 and pg 84.

Why is the average written date in the historical period on page 136 not 6months prior to the average earned date?

[Vorian Atreides]

I think it's an unfortunate use of the term "earned".

Remember the focus is on expenses, not premium. Expenses are "earned" the moment that they're spent; it's not over time.

So, of all expenses spent in calendar year 20X4, the average spend date (which the text is using the term "earned date") is mid-year: 7/1/20X4.

The average written date isn't any different that in other contexts; so for the calendar year, it's 7/1/20X4, the same as the average spend date for expenses.

[Vorian Atreides]

Note that the implicit assumption here is that expenses are incurred (i.e., spent) uniformly over the time period in question.

[bermi]

Note that the implicit assumption here is that expenses are incurred (i.e., spent) uniformly over the time period in question.

Thanks VA. Much appreciated.

Based on your logic, can we conclude the trend from date for expenses will always be the same regardless of "type of expense", and the trend to date will only change based on whether we consider the expense to be:
- incurred at onset
- incurred evenly throughout the policy period

And obviously trend for (expenses incurred throughout) will be longer duration than (expenses incurred at onset).

[Vorian Atreides]

Good question, and yes.

"Incurred at onset" is directly tied to the assumption of how are policies written distributed (with the additional assumption that the average premium will be pretty much the same throughout).

"Incurred evenly throughout the expense period" has an anologous reasoning.

And your last sentence is not correct. When analyzing expenses, we're not tying them to any particular policy period; it the expenses incurred during the calendar year that is under consideration when assessing expense trend. So the trend to date will still be the same (using the same reasoning to determine the trend from date). (See posts below.)

[bermi]

Good question, and yes.

"Incurred at onset" is directly tied to the assumption of how are policies written distributed (with the additional assumption that the average premium will be pretty much the same throughout).

"Incurred evenly throughout the expense period" has an anologous reasoning.

And your last sentence is not correct. When analyzing expenses, we're not tying them to any particular policy period; it the expenses incurred during the calendar year that is under consideration when assessing expense trend. So the 'trend to' date will still be the same (using the same reasoning to determine the trend from date).

Based on Werner, we are looking at future expenses as premium is written and premium is earned.
1. Our 'trend from' date for CY historical expense data is the same for the 2 types of expense treatment.
2. Our 'trend to' date is based on the avg written date of policies (expenses incurred at beginning) or avg earned date of policies (expenses incurred throughout) in the future policy period.

So...(Avg earned date of policies) > (avg written date of policies) in the future policy period, so trend for CY expense data will always be larger for (expenses incurred throughout) vs (expenses incurred at beginning)...no?

[Vorian Atreides]

I just looked at that section, you are correct in using a different trend to date for expenses. I'll correct my earlier post.

[bermi]

I just looked at that section, you are correct in using a different trend to date for expenses. I'll correct my earlier post.

Ok sounds good. I am still not completely sold on the trend from date - but I've given this enough time. Will come back later if I need to. FORWARD!!

THANKS FOR THE HELP VA!

[Vorian Atreides]

WRT the trend from date, you're using CY information as a proxy for the PY experience period; but you're trending to a PY forecast period.

And for expenses, getting PY data would be impractical (where getting PY losses + ALAE could be practical), so you're looking at CY data about 99 times out of 98 for expense trending.