Classification System for Fire Losses in Homeowners Insurance*

Proposer/Liason: Daniel Padilha

AAA Michigan is the largest personal lines property and casualty insurer in Michigan and provides homeowners insurance to about one tenth of all homes in the state. A major component of a homeowners insurance policy is a built-in coverage for fire losses. This built-in coverage provides protection against fire losses to the structure of a home as well as items inside the home.

The amount of loss due to fire is not homogenous throughout the state: a brick house next to a fire station would probably experience a much lesser loss given the occurrence of a fire than a wood-frame home in a remote rural area. At the same time, the exposure to fire loss is also not homogeneous: data suggests that many more fires occur in urban areas (even if there are many nearby fire stations) than in rural areas (where the closest fire station may be several miles away). Hence, several factors may influence the frequency and the severity of fire losses in different regions of the state.

For many years most insurance companies have assessed their exposure to fire losses based on a risk classification system developed by the Insurance Service Office (ISO). The ISO’s system assigns a Protection Class code (PC) to each city, village, and township in the country. The codes are based on two main criteria. First and foremost, most communities are assigned a PC code based on a series of standard engineering surveys that include accounting for the proximity of the community to a responding fire station, how well outfitted the fire station is with modern fire-fighting and communications equipment, how well trained the personnel is, and how adequate the water pressure of the public water system is to combat a fire. However, cities with over 250,000 inhabitants are designated “statistically rated” communities. These can be assigned PC codes based on fire loss experience in the city by itself, without regard to engineering surveys.

The problem is loss data has shown that the ISO system can be fallible. In fact, many companies use in-house data to make customized modifications to the ISO system. In this project, you are asked to come up with a risk classification system for fire loss and fire exposure in Michigan. The format of the classification system and the variables used to determine classes are open. Your classification system may take both loss and exposure into account, or you may have a classification grid to account for losses and exposure separately. We will provide a database of sample fire losses in each region in Michigan. We will also suggest some research papers (including the ISO study) on this topic, but you are free to develop your system from any theoretical and experimental backbone you find applicable.

*Summary prepared by Daniel Padilha, Actuarial Analyst.

Top of Page