Ludwig question

[Examinator]

In reading this paper, Ludwig states that homoeowners is a fairly homogeneous line of business (more so than commercial property, to which he's transitioning), and that the distribution of the ratio of loss amounts to the total insured value is independent of the size of the risk.

Wasn't there a paper on Part 5 (Krakowski or Anderson or Burger or somebody) that said the opposite: that the distribution of loss amount to total insured value did vary by amount of insurance? For instance, low value homes tended to have more total losses than did large homes. The underlying reason for the previous statement was mainly due to fire losses, I believe, upon which this distribution is based, according to both authors. Has anyone else wondered about this?

[laurisssa]

I think what you are remembering from exam 5 is that underinsured homes have more total losses than fully insured homes.

[KidCA]

Actually, I believe you were thinking of Anderson's comparison of size of losses for small properties v. large properties. It makes intuitive sense that small properties burn up quickly and often completely compared to more sizable homes. But I also believe this article recommend that homes were grouped into narrow ranges of property values to determine size of loss distributions. This step makes the group more homogeneous.

As for Ludwig, I think he's saying homogeneos groups of homes have similar size-of-loss distributions. Therefore you can model as a % of insured value. I think that Salzmann's insured values only ranged from $10k-25k, this range is probably narrow enough to keep size of loss distributions constant.

[Examinator]

Actually, I believe you were thinking of Anderson's comparison of size of losses for small properties v. large properties. It makes intuitive sense that small properties burn up quickly and often completely compared to more sizable homes. But I also believe this article recommend that homes were grouped into narrow ranges of property values to determine size of loss distributions. This step makes the group more homogeneous.

As for Ludwig, I think he's saying homogeneos groups of homes have similar size-of-loss distributions. Therefore you can model as a % of insured value. I think that Salzmann's insured values only ranged from $10k-25k, this range is probably narrow enough to keep size of loss distributions constant.That seems to be pretty consistent with the article. I read Feldblum's discussion of this paper (he sure likes to get his hands on everything) and he reiterated the homogeneity requirement that needs to be present in order for such an assumption to be reasonable. Thanks for the insights.