VICA 2018 @ Actuarial Media Center
Welcome
Login / Register

Reserving 4.0 – a vision of real-time reserving

Sorry, this video is only available to registered users.

Please sign in below or register to view this video.

Thanks! Share it with your friends!

URL

You disliked this video. Thanks for the feedback!

Sorry, only registred users can create playlists.
URL

Added , Speaker Marcel Wiedemann (Hochschule Esslingen), ICA2018, in ASTIN
275 Views

Description

Speaker: Marcel Wiedemann

What if claims reserving could be done at the speed of light? We are proposing exactly that! Reserving 4.0 – our vision of real-time reserving, making it faster, more detailed as well as enhancing quality. But, why are we proposing such a bold vision?

Today, reserving is mostly based on aggregated data and methods such as chain ladder. Aggregated methods, however, only yield aggregated results making it hard to understand what is really going on in a claims portfolio. It is for instance almost impossible to understand the major drivers such as inflation, legal changes, seasonal effects, policyholder behaviour, …  For this, analysis at the level of single claims is essential and we believe that this is the way forward in today’s world full of data with modern big data techniques only waiting to explore it.

Our vision of Reserving 4.0 combines real-time statistical models for single claims together with their implementation into the claims system. Once a claim is created by the claims handler, the available information can be used to automatically generate a detailed best estimate reserve, ready for immediate use for claims steering and business management purposes.

In our talk we show how we are currently working towards making our vision reality for a large German nonlife portfolio. We describe the steps taken from analysing claims handling processes and collecting necessary data via samples to developing statistical models for single claims and implementing them into the claims system. We also discuss various practical problems we were faced with, especially in getting adequate data for modelling. Moreover, we share interesting insights and results.

Tags: Data Science

Post your comment

Sign in or sign up to post comments.

Comments

Be the first to comment