 |
|
|
|
|
|
|
|
|
|
| Claims Predictive Model Implementation: |
|
|
|
Bayesian Networks help auto insurer reduce average severity by identifying overwritten auto repair estimates. |
|
|
| |
|
|
| Overview |
|
|
|
- Overwriting of auto repair estimates translates into millions of dollars in loss leakage every year.
- Inconsistent and random reinspections do little to reduce the problem.
|
|
|
|
|
|
|
|
Our Approach |
|
|
|
- Interviewed company experts to determine major sources of overwriting, as well as gaps in reinspection process.
- Used Internal and External data to develop CPMS – Claims Predictive Modeling System, which relies on Bayesian Networks to evaluate disparate sources of information consistently and automatically.
- The model captures the experts’ common sense in determining the likelihood that an estimate is overwritten. This is used to allocate resources to claims with the highest likelihood of overwritten.
- CPMS is currently in production and being rolled out nationally.
|
|
|
|
|
|
|
|
|
Outcome |
|
|
|
|
- Claims reinspectors are allocated more efficiently, by focusing on estimates that have a higher likelihood of overwriting.
- Over time, this is changing the behavior of appraisers and repair facilities' in the field, and reducing average severity.
|
|
|
|
|
|
|
|
|
|
|
|
|
| Copyright - TNC Management Group LLC © 2009 |
45 Park Place South, #301, Morristown, NJ 07960 | 1-888-821-1115 |
|
|
|
|
|
|
|
