Predictive Modeling for Small Commercial Risks CAS PREDICTIVE

Predictive Modeling for Small Commercial Risks CAS PREDICTIVE MODELING SEMINAR Beth Fitzgerald ISO October 2006

Agenda • Definition of Risks • Evolution/Challenges of Underwriting • Use of Statistical Modeling • Implementation of Predictive Models • Relativity Analysis

Small Commercial Risks • Size – Area – Gross sales – Low premium • Type of risk – Office, apartments/condominiums, retail, service – Contractors, restaurants, motels, self-storage facilities – Light manufacturing • Rating – CPP vs. BOP

TYPES OF SMALL COMMERICAL RISKS

Growth in Small Businesses Source: Office of Advocacy, U. S. Small Business Administration

Small Business Underwriting Challenges • Low average premium – Doesn’t warrant expensive hands-on underwriting. • Underwritten more as a commodity • Experienced underwriters focused on larger accounts

Underwriting Small Commercial Risks • Establish underwriting guidelines for type and size of risk • Review application information – Numbers of years in business – Financial information – Location information – Building characteristics

Market Research – What the market says it needs • Fast and consistent small business underwriting process • Take advantage of technology • Add intelligence to the policy writing process

What Makes Statistical Modeling Possible? • Advanced computer capabilities – Processing – Data access • Advanced statistical data mining tools

Uses of Statistical Modeling • Scoring of small commercial risks – Improve loss predictability of risks – Increase accuracy of pricing decisions – Cost effective, consistent underwriting • Improve manual rating of risks

Development of Scoring Models • Analyze historical policy and loss data • Link policy and loss data with internal & external data: – Business operational & financial data – Location data – demographic, weather – Other – building, agency • Use statistical data mining software and techniques

Modeling Process Data Linking Data Gathering Data Cleansing Evaluation Analyze Variables Business Knowledge Determine Predictive Variables Modeling

Scoring vs. Rating Manual • Evaluation of scoring variables relative to rating factors • More refined detail than rating manual • Factors not included in rating manual

Modeling Issues with Small Commercial Risks • Less homogeneous risks than with personal lines risks • Variable selection varies by peril and type of risk • Business operational and financial data not always available

Implementation of Model Solution focus/usage: • • • Suitability of risk for underwriting decision Source for additional pricing factors Consistency in underwriting/pricing decisions Compliance with regulations based on implementation decision Consider model alone or model with other information available from application

Implementation of Model Workflows: • Underwriting – – • New Business Renewal business Rating – – Pricing Coverage Adjustment

Business Implementation of Model • Strategic Plan - need management involvement • Prepare Announcement/Training Material for Internal & External Customers • Coordinate Implementation • Monitor Feedback/Adjust Implementation

Benefits of Scoring Model • Reduction of underwriting expense through automated scoring process efficiencies • Fast, cost-effective tool to help you determine which risks to insure • More accurate pricing decisions • Expansion into new markets

Risks of Not Scoring • Lost market share • Greater risk of adverse selection

Use of Statistical Modeling in Manual Rating • Improve rating relativities of current rating factors • Add new rating factor to manual using a multi-variate statistical model

Amount of Insurance Relativities • Amount of Insurance identified as important variable in BOP Scoring analysis • Decision to include as variable in manual and not in scoring model

Multivariate Analysis for Amount of Insurance Relativities • Variables used for Property – Occupancy – Sprinklered–rating identifier – Protection – Construction

Predictive Modeling for Small Commercial Risks • Increased implementation of models in underwriting/pricing of risks • Account view vs. individual line of business view – BOP, CPP, CA, WC • Set of risk component variables in addition to overall score • Additional data sources • Refinement in manual rating