The New ABCs of GIS

A Conversation with Bill Raichle

Bill Raichle is vice president, Risk Decision Services (RDS) at ISO, part of the Verisk Analytics risk assessment business segment. He has product management responsibility for ISO's ProMetrixTM suite of commercial property products and RDS's product development, marketing, engineering, GIS strategy, and financial planning units. Bill recently sat down with ISO Review to share his thoughts on GIS benefits, advancements, and future uses.

ISO Review: How can GIS be defined from an insurer's ­perspective?

Bill Raichle: A geographic information system (GIS) is a computer system that can store, analyze, and output data, reports, and visual maps of geographically referenced information. The information in these systems can be the policy addresses in a book of business, related underwriting and rating information associated with each policy, coastlines throughout the country, windpools or other regulatory boundaries, natural catastrophe-related data, U.S. census demographics, rating territory boundaries, fire-protection classes, wildfire fuels, business locations, road networks, and so forth. A robust GIS can aggregate diverse types of information into a coherent resource available for creative analytic assessment.

GIS technology itself isn't useful to insurers without industry-specific applications. Out of the box, a GIS is about as exciting as a blank spreadsheet. But when insurers augment their internal policy and business data with current high-quality, real-world spatial risk and market data, new business capabilities and solutions become possible.

ISO Review: How have insurers traditionally used GIS, and what benefits have they achieved?

Raichle: Property/casualty insurers have realized tremendous value from GIS technology since the mid-1990s. We estimate that insurers who have built GIS-related services into their standard business processes represent more than 90 percent of direct written premium for property lines of business.

Most insurers now use GIS-based information for accurate underwriting and rating at the point of sale. GIS helps insurers assign rating territories and public protection and building code effectiveness classifications. It also aids in assessing distance to coasts and wildfire exposure, flood zone determination, and more. And for insurers who have not yet adopted a GIS, it offers immediate hard-dollar value.

Insurers have also relied on GIS for risk concentration assessment of their catastrophe risk management activities, and they have access to powerful damageability models to assess exposure to hurricanes, severe storms, and earthquakes. Those models require accurate geospatial and environmental inputs that only an appropriately populated GIS can deliver. The sophisticated geographic processing involved in those applications produces real value. The better products on the market have been widely adopted.

I've seen many insurers stem millions of dollars in premium leakage each year by using GIS to identify accurate rating variables. GIS assessment has also become standard procedure for insurers and reinsurers in negotiating catastrophe treaties.

ISO Review: What advancements has GIS technology experienced recently, and where is it headed?

Raichle: I'll return to the spreadsheet analogy again. The underlying GIS platforms alone are as mature as Excel. There will always be updated versions of software, with some new bells and whistles, but the core functionality is well established and readily available. The real excitement and the most beneficial advancements are coming from better data — and the flight to quality in GIS data has accelerated.

Base-level GIS data — the street files that drive geocoding— is improving all the time, and smart users are running files from multiple vendors. Tele Atlas's "point" locations, county parcel information, and NAVTEQ's large breadth of roads can be combined for true high-octane results, and we've been successful in doing so for years at ISO. High-resolution data continues to increase in availability and decrease in price as advances in remote sensing technology and processing capacity proliferate.

Overall, geocoding technology is advancing well with regard to positional accuracy. The next phase is to redouble efforts on the timeliness of updates.

The depth and quality of risk assessment data is also at an all-time high. For example, in our labs, we use GIS analysis and remote sensing capabilities to construct our own multifactor wildfire hazard databases that aggregate a number of diverse data layers. Wildfire hazard assessment requires combined analysis of current road network, fuels distribution, and terrain slope data. Those databases, along with our robust geocoding capability, typically identify more than 95 percent of the affected buildings in recent California wildfires as being exposed to the hazard. And consider the fact that insurers now have access to databases of fire hydrants, suction points, and hauled water distribution for almost 90 percent of the U.S. population. Such data-collection feats would have been unthinkable ten years ago. But these are now well-maintained industry data assets that are at the disposal of insurers today.

ISO Review: How can insurers use GIS to develop useful analytics and predictive models?

Raichle: A number of insurers are expanding GIS deployment among their analysts for use in marketing, building risk models, and better understanding their current books in relation to the larger market. For example, a writer of high-value homes in suburban Georgia counties asks, "What percentage of homes in excess of $400,000 in Coverage A do we currently write in the state, and by county?" and "Are our policies generally more risk-prone than our target-market population?" The best work here is yet to be done.

A GIS is also an efficient engine for the manufacture of independent-variable candidates for insurer modelers. For example, a large homeowners writer uses data on community investment in firefighting communications and response equipment to predict burglary loss — and it works well. It works because communities that choose to invest in upgrading fire-mitigation capability might also be more likely to invest in protecting their citizens from other perils, like crime.

The creation and analysis of geospatial information spurs other innovations. Let's look at two personal lines examples:

First, in personal auto insurance, new data features about where driving actually occurs are being created at a refined geographic level. The majority of accidents occur close to home, so it seems natural to investigate complex geographic variables that add value to existing loss models. From an estimated loss cost perspective, this analysis examines the interactive effects of traffic environments (including numbers and types of businesses in an area, traffic density, driving patterns, and public transportation usage), the natural environment (such as weather and topographical characteristics), and other factors (such as claim frequency and severity trends by state).

Similarly, for homeowners insurance, GIS can combine industry statistical data and loss cost information, property and hazard information (such as fire-protection classifications and distance to coasts), U.S. census demographics, weather norms, and business location data. The result is powerful analytics that uses predictive modeling to examine hundreds of indicators that can affect homeowners risks and predict the noncatastrophe portion of expected losses at the policy level by peril.

The industry should expect to find geospatial innovations in the future that apply directly to many lines of insurance.

ISO Review: How costly is it to build and implement a robust GIS in an insurer environment?

Raichle: Like everything else in our business, 80 percent of the cost and value lies in the insurance-specific data content. Twenty percent or less of the cost is hardware, software, and basic GIS databases. And just like everything else, data acquisition is a garbage-in-garbage-out proposition. You can populate a GIS for $10,000 or spend millions — and some have. But the only way to assess such an investment is to become educated about what is possible, be specific as to what value you intend to achieve, and make a business case. A GIS investment has to hold up to real financial analysis. And, of course, industry vendors will continue using the economies-of-scale approach to develop some interesting applications, so insurers won't have to reinvent every wheel.

ISO Review: Where are insurers headed with GIS technology in the next five years and beyond?

Raichle: The real value in a technology comes less from the answers it provides than the new questions it allows you to consider. And that is certainly true of GIS.

As competition continues to thrive in the industry, insurers are asking increasingly complex questions about the role geography can play in gaining competitive advantage: How does spatial location and proximity influence new business acquisition, retention, and profitability? Where are the consistently profitable hot spots in our books of business today? How can we best tailor our products by customer segment and geography? How can we understand the possible effects of climate change on our books of business? How can our statisticians and actuaries best reflect geographic variables in our pricing and underwriting models?

As GIS becomes a standard tool, the "gee-whiz" buzz about base maps and imagery will fade, and users will become more demanding about proving returns on their investments. In this maturing GIS technological environment, asking the right questions about the use of geographic variables in your business is the path to solid returns. That's good business sense for any technology, and it's the long-term future for GIS.