CEIOPS highlights the need for proper data handling

Some time ago, I shared my vision of professionally integrated actuarial systems, in a dada talk at the PillarOne conference in Brussels. Any change in topics since then? In my experience, the quadruple "Data, Actuarial, Documentation and Accounting" still is often not as balanced as one might expect it to be. So I will blog through my latest experience with these four parts of the quadruple during the next weeks.

Starting with "D" as standing for Data.

A friend of mine, responsible for internally reviewing reserve estimates for several group companies of a global insurance company, sounds quite nervous every three months during closing time. And I guess, this is not due to tight deadlines only, but rather to several emails in his inbox advising on different numbers not really fitting into the overall picture as provided through a narrative in - just another - email.

"Yeah, we made some changes."
"Oops, we forgot to include the special exclusions."

I guess a proper data store would help with the first data-D in dada. This is fully in line with what the industry expects: A recent publication from CEIOPS highlights the need for proper data handling by asking three questions to insurance companies:

• Is your data appropriate?
• Is your data complete?
  
• Is your data accurate?

Having a proper data store in an insurance company would answer these questions easily: all risk related data is stored in a dedicated data store, sitting on a server under the responsibility of the IT department. Regular backups included; seamless login and access rights/restrictions well-defined; all existing risk software pulling their input from this data store and pushing results back to it through secure data loads (one of them, of course, RiskAnalytics); risk reporting relying on this single-source-of-truth; and so on like global instant access through the company's intranet.

Getting back to the three questions above as they address some very important issues in relating to handling data. If you answer these three positively, you most probable will not have problems in answering others such as the famous CEAVOP: Completeness – Existence – Accuracy – Valuation – Ownership – Presentation. If you answer only one of them negatively, I doubt you will want to further rely on your current "data-solution" in your risk management.

Markus M.