A City of Two ReTails 2003 Fall CAS

A City of Two (Re)Tails 2003 Fall CAS Meeting November 11, 2003 Robert J. Walling

The City of Toningbloom has 2 Hardware Stores… Soft Hardware Sucha Tools

Different, but eerily the same… l Both are: – – – l Joisted Masonry Protection Class 6 Same Coverages and Amounts of Insurance In fact, the properties have identical “traditional” risk characteristics.

They both went insurance shopping…

The quotes came back… Insurer Farm States Soft Quote $1, 900 Sucha Quote $1, 700 Inilli Mutual $1, 800 $1, 900 OSIns. Co $2, 100 Yankees Ins Co. $1, 500 $2, 400

Why? Elementary my Dear Watson…

“Who” Issues, Not “What” Issues l Look at data off the application that is not rated: l l l Percent Occupied Years in Business Years of Same Mgt. Updated Systems Alarms Sole Occupancy Computer Back Ups Hours of Operation Franchise? Safety Program Employee Mix (Full Time, Leased, etc. )

How do companies address these factors?

Traditionally schedule credits and rating tiers? Company Tiering Factor Schedule Percent of Manual Max/Min Barbershop I. C. 1. 25 +40% 175% Barbershop I. C. 1. 25 -40% 75% Vanilla I. C. 1. 00 +40% 140% Vanilla I. C. 1. 00 -40% 60% BTA I. C. 0. 85 +40% 119% BTA I. C. 0. 85 -40% 51% TPet I. C. 0. 70 +40% 98% TPet I. C. 0. 70 -40% 42% THE HIGHEST NET RATE IS OVER FOUR TIMES THE LOWEST!!

Companies are moving to Underwriting Scorecards Using GLM

Applications of GLM for BOP Pricing Enhancements l l Revise Class Factors Revise/Enhance Territories – l l May have impact similar to Homeowners on Protection Create more sound AOI curve Develop Underwriting Scorecard Incorporating – – Credit scores Other “who” characteristics (especially those already required on the application)

The GLM Approach l l l Capture by Policy and by Claim Experience Append Credit Variables and Application Data Develop Frequency and Severity Models – l Consider Relevant Interactions Combine Models to Develop Pure Premiums and Tiering Plan/Scorecard Elements Simultaneously

Underwriting Scorecard Example

Underwriting Scorecard Example

Underwriting Scorecard Example

Issues of Developing BOP U/W Scorecard l l Concerns Over Use of Credit “No Hits” Capture of Application Data Volume of Data for Certain Classes, Territories, etc.

Benefits of Using GLM for BOP Pricing Enhancements l l Reduce Reliance on Underwriting Discretion Improves Predictive Accuracy Creates Adverse Selection for Competitors Reflects Interactions with Between Rating Factors

Credit’s Problem - Interactions

“There’s always a greater fool” Insurer Farm States Soft Quote $1, 900 Sucha Quote $1, 700 Inilli Mutual $1, 800 $1, 900 OSIns. Co $2, 100 Yankees Ins Co. $1, 500 $2, 400 Naïve Capital $2, 720 $1, 350

Adverse Selection Insurer Soft Quote Soft Loss Sucha Ratio Quote Farm States $1, 900 51% $1, 700 Sucha Loss Ratio 92% Inilli Mutual $1, 800 54% $1, 900 82% OSIns. Co $2, 100 46% $2, 100 74% Yankees Ins Co. $1, 500 65% $2, 400 65%

The Impact of Credit and Other Factors May Vary by Class

Underwriting Scorecards Reflecting Interactions l Multivariate analysis allows the modeling of interactions and modern policy management systems facilitate the implementation of more complex tiering systems

Parting Thoughts l Where there is no vision, the people perish. – Proverbs 29: 18 The data’s ready, The technology’s ready, ARE YOU READY? ? ?