COMMERCIAL INSURANCE SCORING SYSTEM AND METHOD

• US 20120271659A1
• Filed: 02/15/2012
• Published: 10/25/2012
• Est. Priority Date: 10/23/2000
• Status: Active Grant

• Current Assignee: Deloitte Development LLC
• Original Assignee: Deloitte & Touche LLP (Deloitte & Touche)
• Inventors: Zizzamia, Frank M., Wu, Cheng-Sheng Peter, Carrier, Matthew R., Lucker, John R., Tocci, Dominic A.
• Granted Patent: US 8,655,687 B2
• Time in Patent Office: Days
• US Class Current: 705/4
• CPC Class Codes: G06Q 10/0635 Risk analysis of enterprise or organisation activities
• CPC Class Codes: G06Q 40/02 Banking, e.g. interest calculation or account maintenance credit or loans G06Q40/03
• CPC Class Codes: G06Q 40/03 Credit; Loans; Processing thereof
• CPC Class Codes: G06Q 40/08 Insurance

Abstract

A quantitative system and method that employs data sources external to an insurance company to generate a statistical model that may be used to more accurately and consistently predict commercial insurance profitability (the “predictive statistical model”). The system and method are able to predict individual commercial insurance policyholder profitability on a prospective basis regardless of the internal data and business practices of a particular insurance company

Claims

1. A method for predicting the profitability of a commercial insurance policy, said method comprising:

obtaining policyholder data including historical premium and loss data;

obtaining external data;

identifying one or more variables in at least one of the policyholder and external data,

creating by a computer one or more derived variables from at least one of said policyholder data, external data and any identified variables;

evaluating by the computer, the identified variables and at least one of the derived variables against the policyholder data;

identifying a set of variables predictive of the insurance policy'"'"'s profitability based on the evaluation of at least one of the identified variables and the derived variables against the policyholder data; and

creating an individually weighted multivariate statistical model based on said set of predictive variables, and an associated score, wherein said score is expressed as a sum of products, each of said products being a coefficient multiplied by a variable taken to a power, said coefficients generated mathematically by operating on said set of predictive variables, and

using said score to predict the profitability of an insurance policy.

2. The method of claim 1, wherein said score is a function of any plurality of said predictive variables.

3. The method of claim 1, wherein said identified variables can be explicit or implicit in the policyholder data and/or external data.

4. The method of claim 1, further comprising creating individual records in a database for each policyholder and populating each individual record with the policyholder data, including, to the extent available, business name, address and zip code for each policyholder and any relevant identified variables.

5. The method of claim 1, further normalizing at least one of the policyholder data and the external data in the database.

6. The method of claim 1, wherein said normalizing said policyholder data includes premium manualization, loss trending and loss capping.

7. The method of claim 1 wherein the external data includes data relating to at least one of geographic factors, business stability and weather patterns.

8. The method of claim 1 wherein said identifying a set of predictive variables further comprises examining a plurality of the identified variables and the derived variables for cross-correlation against one another in order to eliminate repetitive variables.

9. The method of claim 1 further comprising dividing the data into a training data set for developing the statistical model, a testing data set for refining the statistical model and a validation data set for evaluating the predictiveness of the statistical model.

10. The method of claim 1 wherein said identifying a set of variables predictive of the insurance policy'"'"'s profitability further includes normalizing the policyholder data, calculating for each policyholder a loss ratio based on normalized policyholder data, defining a subgroup from the policyholder data, calculating a cumulative loss ratio for the subgroup and performing a statistical analysis to identify statistical relationships between individual external and derived variables and the cumulative loss ratio for the subgroup.

11. The method of claim 1 wherein the statistical model is created using multivariate methods to produce coefficients for each of the set of predictive variables, wherein the coefficients represent the contribution of the each of the set of predictive variables to an overall score.

12. A program storage device readable by a machine, said program storage device tangibly embodying a program of instructions executable by the machine to perform a method comprising:

obtaining policyholder data including historical premium and loss data;

obtaining external data including at least one of business level data, personal data and household demographic data,

identifying one or more variables in at least one of the policyholder and external data,

creating by a computer one or more derived variables from at least one of said policyholder data, external data and any identified variables;

evaluating by the computer, the identified variables and at least one of the derived variables against the policyholder data;

identifying a set of variables predictive of the insurance policy'"'"'s profitability based on the evaluation of at least one of the identified variables and the derived variables against the policyholder data; and

creating an individually weighted multivariate statistical model based on said set of predictive variables, and an associated score, wherein said score is expressed as a sum of products, each of said products being a coefficient multiplied by a variable taken to a power, said coefficients generated mathematically by operating on said set of predictive variables, and

using said score to predict the profitability of an insurance policy.

13. The program storage device of claim 12, wherein said score is a function of any plurality of said predictive variables.