Inflating the threshold in RI pricing

[Oleksii Kovalchuk]

Hello,

on page 22-23 of Chapter 20 the problem of missing claims for earlier years in case of constant reporting threshold is discussed. I seem to grasp the issue, but I am not sure to understand the remedy from Core Reading. It is said that:

"to overcome this, the reinsurer will need to restate the threshold for all but the most recent years in the data. Using the same claims inflation rates as were used to trend the losses, the reinsurer will also inflate the threshold year by year, so that its real value is preserved and any untended losses that never exceeded the inflated threshold for the appropriate year are discarded."

If we restate threshold for past years to express it in real terms and then compare it to untended losses, should we not "deflate" it rather than inflate? And even if that what is actually meant, how would it solve the issue if losses that would have been above such threshold are still missing from data? Or here we mean some other threshold, e.g. retention level rather than reporting threshold?

I feel I am missing some basic point but cannot get what...

 

[Ian Senator]

Don't worry, your understanding is fine. The Core Reading is a bit confusing because it's talking about inflating the threshold forwards (ie after you've deflated it back to the oldest year). As long as you realise that you use a lower threshold the further you go back in time, in order to get a true representation of the number of losses likely to hit the layer, you've understood the issue.