He might have been pleased, but I was depressed. After all of the years of traveling cross-country by those like me, preaching the Gospel of points, lines and polygons, I thought everyone got it by now. I was dead wrong. I know that the insurance industry lags behind most industries in the adoption of Geographic Information System (GIS)-based solutions. And that getting them to make changes for their own good is a Herculean task. But, ZIP Code tables?
For those of you whose knowledge of insurance is limited to the fact that you have some, a rating territory is one of the factors that determines how much you pay for auto, home and other types of insurance. Actuaries use loss data from previous years to determine areas where similar loss patterns exist for similar types of applicants. These areas are then grouped together to make rating territories.
The territory boundaries are described by long flowing sentences that made your head hurt and your ears bleed. Six people could interpret them at least eight different ways. It is no wonder that so many applicants are placed in the wrong rating territories. Sometimes that happens on purpose when agents, anxious to close the sale, pick the nearest territory with a lower rate.
Then came the modern revolution. Why dont we build tables of ZIP Codes that approximate the rating territories? Everyone knows his or her ZIP Code. Don t they? (This is the part where I start to cry.)
The truth is that some people dont know their correct ZIP Code, especially the ones moving into a new neighborhood. These are the same folks that the insurance agents do everything possible to attract.
And theres the problem with ZIP Codes themselves. The U.S. Postal Service (USPS) has the audacity to change them for its own convenience without ever asking for permission from the insurers. ZIP Codes were designed to help speed the delivery of mail. They change all the time as population density shifts. For other purposes, they make no sense at all.
So, if you produce a table of ZIP Codes today, it begins to degrade tomorrow. This creates job security for someone who will attempt to keep it up to date. But, it is a losing proposition, and a bad idea. ZIP Code tables should not exist for any reason that I can think of. Any questions on where I stand on this point? Anybody who has made the mental leap to GIS-based thinking knows how to solve this problem. First, digitize the territories into polygons. Then, standardize the addresses to USPS standards and geocode them. Have your GIS program figure out which polygon (rating territory) the addresses are in and give you the answer. The USPS can change all the ZIP Codes daily, and you will still be rating your policies correctly. It is pure GIS101.
Im going to pack my suitcase again. Apparently, there is still much work to do in spreading the GIS Gospel. See you on the road.
Here's a related letter to the editor.
