This is the process of using predictive models to give insights into the day-to-day underwriting processes of a life insurer. For example, to determine the profile of the client beforehand and determine which people are fast tracked and those that require a medical report
Examples of predictive models we have are Models that use statistics to score the risk profiles of potential clients and provide insights as to which clients require further investigation, e.g medical checkup. With such models, we can now require less people to go through the rigorous process of underwriting & verification hence improving the sale process. This predictive models can be automated in the IT system.
The model can be used in life assurance for several functions e.g.
- Agent selection(selection of productive agents
- Customer segmentation by determining which customers will buy life insurance
- Cross selling where they could determine which existing customer can purchase another product
- Price optimization- different prices for different channels
- Risk selection – risk scoring, ordering underwriting requirements
- Detection of fraud from over insurance or anti-selection
- Pricing
- Reserving in the insurance
Predictive models can be developed as follows
- Data Mining – Establish Patterns, Collect data, clean data and assign data distribution
- Logic & Algorithm – Develop decision trees & identify factors and predictors
- Build Model (can be repetitive) – Build, Test & Calibrate
- Validate
- Implement & Document
- Monitor and Recalibrate
Popular predictive models that exist are;
- Decision Trees
- Regression Trees
- Cox Model
- Generalized Linear Model
- Logistic Regression
- Regression Spline
- Neural Networks
- K-Nearest Neighbour
The advantages of predictive models are;
- Prediction- Customers are happy if the sale process is shortened or the sale is warmer (selling to a client already looking for a particular product)
- Some prediction models require minimal statistical knowledge – neural nets
- Various statistical methods available for prediction models
- Usage of already collected data to improve business process – insurers with rich history, strong data integrity can leverage – perfect for online business
The disadvantages we have with predictive models
- The model may be wrong
- If not checked/updated/calibrated regularly with recent data
- Overfitting/wrong predictors
- May not make sense (common sense)
- Black box – nobody knows what is inside it
- May depend on modeller (biased by perceptions)
- Requires IT infrastructure, data (lots of it) and human capital
