PDA

View Full Version : A couple premium trend questions


asile
02-19-2012, 10:18 PM
1. When calculating premium trend you generally use on level written premium per exposure. Is exposure trend usually included in the on level adjustment? If not, wouldn't it be improper to apply an exposure trend too because we would have caught it in the premium trend?

2. Werner page 82 gives an example about premium trend would include indexed homeowners policies. Ok, but then would you not want a buildings/contents exposure trend too? (assume we are handling the liability separately).


Thanks just trying to make sure I understand and don't double count.

Vorian Atreides
02-20-2012, 12:51 AM
1. When measuring premium trend, you're looking at average premium per exposure (which might be "per policy"), and this would be equal to "total on-level premium" divided by "total exposures". In many cases, exposures don't "grow" over time.

For example, 1 house-year in 1995 = 1 house-year in 2015. Likewise, 1 car-year in 1990 = 1 car-year in 2010. That is, that house or car doesn't get any bigger (or smaller) over time.

However, suppose for CGL, the exposure basis is gross sales receipts. Here, $100k sales in 2005 <> $100k sales in 2010. It is quite possible that the propensity for loss for this risk has not changed, and the volume of business for this risk hasn't change (that is, it is selling the same quantity of material in 2010 as it did in 2005), but the gross sales receipts has likely grown to, say, $128K (5% annual inflation). In this case (that is, the case where the exposure basis is inflation sensitive), you will trend exposures first to a common level before calculating the average premium. This way, you can isolate (or separate) the premium trend due to coverage changes (i.e., your mix of business changing due to shifts in distribution of limits, coverage options, deductibles, etc.) from changes due solely to exposure (as the result of inflation).

2. Consider the following quote from page 82:


The actuary may consider examining how premium distributions by individual rating variable have shifted over time. However, this may not always be practical or conclusive. Such distributional data may not be readily available, or the actuary may find that several variables have experienced small premium shifts and the compound effect is difficult to quantify. Consequently, the analysis usually focuses on measuring all premium shifts simultaneously.

(emphasis added)


There is usually mutliple influences going on, so it's usually more practical to measure all the distributional shifts simultaneously (that is, use house-years as the exposure basis rather than amount of insurance (even if that's "more accurate")).

And based on ISO forms, contents coverage are simply a matter of a percentage of the dwelling amount of insurance (so there's no need to do a separate analysis for contents). And since house-years would be the likely exposure base for homeowners liability (and it's generally not rated "separately", that is, you pay one premium for property and liability coverage, with a charge to increase the liability limit), there's no need to analyze liability separately in this case--that is, there's a gain in efficiency with no (material) loss on the overall analysis.

asile
02-20-2012, 01:23 AM
1. When measuring premium trend, you're looking at average premium per exposure (which might be "per policy"), and this would be equal to "total on-level premium" divided by "total exposures". In many cases, exposures don't "grow" over time.

For example, 1 house-year in 1995 = 1 house-year in 2015. Likewise, 1 car-year in 1990 = 1 car-year in 2010. That is, that house or car doesn't get any bigger (or smaller) over time.

However, suppose for CGL, the exposure basis is gross sales receipts. Here, $100k sales in 2005 <> $100k sales in 2010. It is quite possible that the propensity for loss for this risk has not changed, and the volume of business for this risk hasn't change (that is, it is selling the same quantity of material in 2010 as it did in 2005), but the gross sales receipts has likely grown to, say, $128K (5% annual inflation). In this case (that is, the case where the exposure basis is inflation sensitive), you will trend exposures first to a common level before calculating the average premium. This way, you can isolate (or separate) the premium trend due to coverage changes (i.e., your mix of business changing due to shifts in distribution of limits, coverage options, deductibles, etc.) from changes due solely to exposure (as the result of inflation).

2. Consider the following quote from page 82:

[/SIZE][/FONT]
[/SIZE]
There is usually mutliple influences going on, so it's usually more practical to measure all the distributional shifts simultaneously (that is, use house-years as the exposure basis rather than amount of insurance (even if that's "more accurate")).

[/FONT]And based on ISO forms, contents coverage are simply a matter of a percentage of the dwelling amount of insurance (so there's no need to do a separate analysis for contents). And since house-years would be the likely exposure base for homeowners liability (and it's generally not rated "separately", that is, you pay one premium for property and liability coverage, with a charge to increase the liability limit), there's no need to analyze liability separately in this case--that is, there's a gain in efficiency with no (material) loss on the overall analysis.

Thanks, CGL example makes perfect sense. I work in comm'l lines so I was thinking homeowners was 100 of TIV like comm'l prop; I can see why house years makes more sense (and then you just pick up the impact of an indexed limit in your premium trend).