1 / 19

Casualty Loss Reserve Seminar 2003 Session #4 Workers’ Compensation Reserving – How and when should you slice the cake?

Casualty Loss Reserve Seminar 2003 Session #4 Workers’ Compensation Reserving – How and when should you slice the cake?. Ezra Robison, FCAS, MAAA 09/08/2003. Today’s agenda:. Failing to slice the cake Implications for triangle technology Could we generalize slicing?. Pursuing relevancy.

rusty
Télécharger la présentation

Casualty Loss Reserve Seminar 2003 Session #4 Workers’ Compensation Reserving – How and when should you slice the cake?

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Casualty Loss Reserve Seminar 2003Session #4Workers’ Compensation Reserving – How and when should you slice the cake? Ezra Robison, FCAS, MAAA 09/08/2003

  2. Today’s agenda: • Failing to slice the cake • Implications for triangle technology • Could we generalize slicing?

  3. Pursuing relevancy • Are actuaries pursuing increasing precision in areas of decreasing relevancy? • Actuarial science is forecasting • Slicing the cake is about forecasting • A better slicer is not decreasingly relevant

  4. Anecdote – failing to slice the cake • All WC • All the same state • 4 separate business units

  5. Moral of the story • I was using one body of data to assign a tail factor to a different body of data • A central issue in considering “how to slice the cake” • Sometimes, there is no alternative • Sometimes, this is done unknowingly

  6. A brief history of technology and triangles

  7. What is the goal? • Optimize the balance between credibility and homogeneity • Systematize what is currently, generally, an ad hoc process

  8. How would we optimize the balance? • Use claim, premium and exposure information at lowest reasonable level • Gather statistics by dimension (slices) • Construct relevancy statistics • Construct credibility statistics • Construct trade-off functions • Find the slicing that optimizes the relevancy and credibility

  9. Are these thoughts relevant to non-primary carriers? • It is not immediately obvious that those who do not own vast claim or sub-claim level detail can pursue this direction immediately • But valuable use of existing data will drive the development of technology throughout the industry • E.G. catastrophe modeling

  10. What is relevancy? • Based loosely on Howard Mahler’s, "An Example of Credibility and Shifting Risk Parameters," PCAS LXXVII, 1990 • The extent to which one body of data is relevant for predicting the future of another body of data • Relevancy is closely tied to the concept of homogeneity • I like it because I find it easier to think about quantifying relevancy

  11. How would we define relevancy? • A formal definition might be based on the percent of policies (premium, exposure or expected loss cost) common to both bodies of data • Could also incorporate claim or sub-claim level detail (e.g. PPO savings)

  12. Relevancy

  13. A graphical representation

  14. How would we measure relevancy over time? • Trend? • Lower trends imply higher homogeneity • Average? • Higher averages imply higher homogeneity • Minimum? • Higher minimums imply higher homogeneity

  15. What do we mean by credibility? • Probably not a formal definition such as Buhlmann or Classical • Rather, a general concept: larger bodies of data are inherently better for forecasting

  16. What would our credibility metric be? • Number of buckets? • Fewer buckets implies higher credibility • Average size of bucket? • Higher average implies higher credibility • Minimum size of bucket? • Higher minimum implies higher credibility

  17. Finding the optimal combination • We may have to settle for subjectivity for now, but possible standards include: • Obtaining a minimum standard for each of credibility and relevance • Maximizing a subjectively derived measure which is a function of both credibility and relevancy • This is similar to maximizing the economic concept of utility

  18. What would we gain by this approach? • Guidance in slicing the cake • Quality control

  19. Remaining issues: • The subjectivity associated with finding the right balance • The measurement of both credibility and relevance are complicated in that they change over time

More Related