data requirements for statistical reserves

Discussion in 'ST3' started by Gareth, Jul 8, 2006.

  1. Gareth

    Gareth Member

    Page 3, chapter 14 (outstanding claims reserves) says "If reserves are to be estimated by statistical methods...the data will need to be far more detailed...comment on how this is done appear elsewhere in this and other chapters of the course".

    I couldn't find much in the chapter about it, which just leaves the data requirements stuff in chapter 10 for premium rating.

    I'm thinking the data requirements should be more or less just the claims data requirements for rating - i.e. detailed policy data is less relevent here?

    Can anyone think of additional data requirements required for reserving above those of rating?
     
  2. Muppet

    Muppet Member

    agree - in fact you probably need less detail for reserving than rating because you will group more things together.

    Bit of an "iffy" bit of core reading. It's comparing against case estimation - when you would probably use more specific info about outstanding claims - but less info overall (ie not historic claims info).

    Detail is touched on again in Chapter 24.
     
  3. parnell

    parnell Member

    I'm pretty sure the "less" aspect reflects a comparison with setting case reserves individually i.e claims technicians setting reserves rather than relative to premium setting.

    Having said that I would add that one is not comparing like with like here in your question - premium setting is prospective whereas reserving is retrospective - when setting reserves on a particular policy segment although the data is grouped - there would be more data available - simply because the period of expsoure has passed and there would be a great deal of knowledge available about the specific cohort year for that piece of business e.g latent claims exposure (could not possible be accurately allowed for at time of writing business).
     
    • claim settlement date
    • date claim was re-opened (if applicable)
    • and.....basically enough info to split the data into whatever groups you want

    The ActEd bit of that page says that just about any piece of info relating to the circumstances of the claim can come in handy. I think this holds true especially for very large claims - any 'soft' info about the claim could be useful to help make a better estimate
     

Share This Page