AAIS licenses “Pretium” software
from Watson Wyatt Worldwide

In actuarial circles, there are no bigger names than Watson Wyatt.

The global consulting firm that rose to prominence through its expertise in pensions and employee benefits is now expanding its presence in property/casualty insurance.

One of its signature products is “Pretium,” a comprehensive predictive modeling software platform that allows actuaries to test and develop predictive factors quickly and cost-effectively.

More than 500 insurers throughout the world use Pretium, about 40 of them in the United States.

And, now, AAIS is one of them.

AAIS recently signed a license with Watson Wyatt Worldwide to utilize Pretium for its initiative to develop rating plans that will rate individual property perils and liability claims, and combine them to produce a final loss cost. The initiative is being undertaken first in the AAIS Homeowners Program, and will later be expanded to other lines.

“Pretium brings multivariate modeling and sophisticated geographical spatial analyses to AAIS,” says Roosevelt Mosley, a consulting actuary for Pinnacle Actuarial Resources, Bloomington, Ill.

“These applications are being used by insurers to help identify profitable segments, fine-tune marketing initiatives, and incorporate customer response modeling,” Mosley says. “Pretium streamlines the analysis process for faster and more efficient modeling.”

As a complement to its license with Watson Wyatt, AAIS has also signed a license with Pinnacle to provide support service for Pretium.

Data

The licensing of Pretium is one aspect of a larger AAIS initiative to utilize predictive analytics on behalf of its member companies, says Paul Baiocchi, president.

According to Baiocchi, AAIS has acquired data on approximately 100 different potential variables that can be tested for predictability of loss.

Nearly 60 categories of data have been acquired from the U.S. Census Bureau, including data for geographic areas on housing, population, transportation, and other categories.

In addition, nearly 40 categories of data on crime and weather conditions throughout the U.S. have been acquired from a firm that provides demographic and industry data to other companies.

“This external information enriches our data resources,” says Baiocchi. “Combined with our robust data base of traditional insurance statistical data, this new information will enable AAIS to respond to companies of all sizes that are seeking more refined information for rating policies.”

“It is central to our mission of helping companies compete,” he adds.

Capability

The role of Pretium in that mission, Baiocchi says, will be to allow AAIS actuaries to efficiently test a series of variables to determine if and how they are uniquely significant in predicting loss.

“Insurance data is often ‘noisy,’ and can present problems for the accurate analysis of risk and behavior,” says James Tanser, global leader of Watson Wyatt’s Pretium and predictive modeling teams.

“Pretium has been designed to cope with the problematic data often found, and will identify and correct several common problems.”

According to Tanser, Pretium operates as a separate program alongside the SAS statistical application, essentially telling the SAS what to do without requiring complex coding.

“While many of the techniques in Pretium are standard for the insurance industry, Pretium includes some unique functionalities not found in rival software,” Tanser says.

“In particular, Watson Wyatt has extended the functionality of SAS by writing two procedures for it that are available only to Pretium users,” he adds. “These allow for faster application of generalized linear modeling and spatial smoothing.”

Rate impact

Once final models and new indications have been developed, Pretium helps analyze the impact they will have on current and potential customers.

“Will any segments receive large premium increases?,” Tanser asks. Are cross-subsidies present in the indications?

“Pretium allows users to understand the impact of rate changes at a policy level,” he says. “This allows managers to fine-tune the final selected rates.”

Understanding customer behavior is increasingly important in pricing, Tanser adds, and many insurance companies are combining models of customer behavior with models of risk, something Pretium can help them do.

“Pretium will allow us to analyze new rating variables and old rating variables to a degree that couldn’t be done before,” says Greg Jaynes, AAIS chief actuary.