## The Name of the Game

We have written frequently on the importance of deduplication for mortality modelling.  In a mortality- or longevity-related transaction, it is critical that the risk-taker performs deduplication when fitting a statistical model to experience data.  The reasons for this range from having a better picture of the total risk per person to ensuring the independence assumption in modelling and avoiding bias.

Despite this, it is not uncommon for the cedant in a reinsurance transaction to omit names or National Insurance numbers in the experience data, at least in the first round of bidding.  Sometimes this information is only provided when the final bid is accepted.  There are two facets to this:

1. The cedant's data-protection officer (DPO) does not want to give out personal data unnecessarily.  The risk of a breach of personal data is a board-level concern — or at least should be! — so the DPO needs a compelling justification to release it.
2. The cedant may not fully understand the role of personal information and its importance in risk analysis and pricing, so the DPO is never given the compelling reason she needs.

It therefore is worth spelling out why personal information can be so important to a reinsurer or insurer wanting to gain a proper understanding of risk.  When pricing a transaction, the risk-taker has to add prudent margins for uncertainty, especially where there is missing data.  In an earlier posting we saw how the parameters of a model can be biased when deduplication is not performed.  Thus, if the cedant in a transaction does not provide the means to deduplicate, such as names or National Insurance numbers, then the reinsurer faces additional uncertainty.  This will lead to an extra margin in any quoted price, and so not sharing such relevant information has a potential cash cost to the cedant.

Enabling reinsurers and insurers to perform deduplication — and thus reduce risk margins — is one reason why it can be in the cedant's interest to share the likes of names and addresses.  A second reason is the pricing of spouses' benefits.  This is a financially significant assumption: adding 10% to the assumed proportion married can easily add 1% to the price of a bulk annuity.  However, if no personal information is available on the pensioners, an insurer or reinsurer will be forced to adopt a fairly conservative assumption as to what spouses might exist, again leading to an extra margin and a cash cost to the cedant.

So, not providing personal data can leave the cedant at a financial disadvantage when trying to get the best price for a reinsurance quotation.  The two examples above will often give the cedant's DPO the compelling reason she needs to authorise the sharing of personal information, at least in the final bidding rounds.  Of course, the DPO will still want to ensure that only the minimum required data fields are shared, that the personal data are secure when shared, and that they are held only as long as is required to assess the risk.

### RECENT POSTS

A short and simple administrative announcement ... In common with ... Read more
The upcoming EU General Data Protection Regulation places focus on ... Read more
Assume we have a random variable, $$X$$, with expected value ... Read more
Stephen Richards is the Managing Director of Longevitas
##### Deduplication in Longevitas
Longevitas users can control all aspects of deduplication — including switching it off — in the Deduplication tab in the Configuration area.  There are ten different deduplication schemes that you can choose to apply, depending on what data you have available.