Property Loss Scales

[Morrison]

Does anybody know the proper way to apply property loss scales when pricing layers? Specifically, I am trying to figure out whether I should apply the loss scale to the entire property portfolio of the insured (all values combined) or individually to the insured value at each location. I guess the correct answer depends on where the scales came from.

However, it seems to me that if my insured has two locations, each worth $100 million for a total insured value of $200 million. If I applied the loss curve to each location individually, it would suggest a premium of $0.

[MNBridge]

I'll assume you are looking at a retention of $100 M.

If one event (Non-CAT) could wipe out both locations then I would consider them to be one location. Assuming this is not the case (which I probably would) then I would say yes the exposure to the layer is 0.

The CAT modeling people will consider the exposure of the locations in an aggregate and reflect a cost in the EP curve.

My 2 cents.

[Morrison]

Yes, I am looking at the first $100 million.

I guess my next question would be, do "Loss Curves" or "% of Premium" curves include Cat Experience or not?

The curves have names like "New First Loss" and "Lloyds Scale". Thanks for you answer.

[joeorez]

There are Proceedings papers by Salzmann in the 60's and Ludwig in the 90's for how their scales are derived. Good luck finding out how any other scales were derived. Certainly the Salzmann (homeowners data) and Ludwig (homeowners, small package data) scales were not derived from properties of the size to develop a 100 million loss on a single risk.

If you are pricing an individual risk cover with a 100 million retention, I don't think any scale has enough data behind it to price that. Talk to the underwriter instead. If you are pricing a catastrophe cover, then use a cat model.