July 3, 2018 | Written by: Patrick Sheridan
Categorized: AI | InsurTech | Risk & Analytics
Share this post:
The first known practices of “insurance” are from hundreds of years BC when neighbors would all come together to rebuild a farmer’s barn lost to fire. Or when the community started to store grain so that they would have food in years of famine. Although this part is not documented, I am sure somewhere along the line that a person said, “I am not rebuilding his barn, this is the third time”, or, “I don’t want his grain in the community store. It is not good.” These unsung somebodies would be the examples of the first underwriters.
Underwriting: Fundamentally necessary
As defined by The Economic Times,
“Underwriting is one of the most important functions in the financial world wherein an individual or an institution undertakes the risk associated with a venture, an investment, or a loan in lieu of a premium. Underwriters are found in banking, insurance, and stock markets.“
Insurance underwriters are professionals who evaluate and analyze the risks of insuring people and assets and establish acceptable premiums for the accepted risks. Underwriters help price life insurance, health insurance, commercial liability insurance, homeowners insurance, etc. Underwriters use computer programs and actuarial data to determine the likelihood and magnitude of claim payouts over the life of the policy. Evaluating an insurer’s risk before the policy period and at renewal is a vital function of an insurance underwriter. In some cases, underwriting is not only to determine “if” to accept the risk, but also as input to accepting at “what premium?”
Insurance underwriting continues to evolve
While the process of underwriting has been around forever and the processing of insurance has become more and more automated, it is still as much of an art as a science. In today’s world, we can see that a potential insured has been cited in multiple previous auto accidents, and we do not want to take them on as a new risk (unless there is acceptable premium for that increased risk). It is not so cut and dry on Main Street America where we have no background on whether a beauty shop will survive there or what the crime rate will be in this new revitalized part of town. Can a towing business survive here? In these situations, underwriters are forced to gather as much information as they can, look at past similar businesses, and use their best judgment. Data considered include business starts and stops in the area, crime statistics, demographics, fire runs, weather patterns historic and future, etc. While it is easier to get these various points of data, the process of analyzing the data takes longer and longer. Years of experience show that the longer it takes to process an insurance application, the more likely the company is to lose that prospective customer. Historically, it has been tracked that the underwriting process can cost as much as 2.1% of Gross Written Premium and take as long as 45 days to fully evaluate. So in addition to being accurate, the underwriter must work under the clock also. While zero customers may manage the risk, it is not a viable business model from a revenue perspective. Time is of the essence for the underwriting process.
Processing large and complex amounts of data
Let’s get back to the available data. Today, the data comes in all forms and we have learned how to glean useful facts from all of it: social media, handwritten police reports, witness statements, written and verbal, accident scene photos, satellite photos, drone videos, and, of course, statistics and textual records. All of these data sources require a different skill level to effectively interpret the information they contain. Combine this complexity with the fact that almost all insurance products require a subject matter expert trained underwriter to know what to look for in all of this data to identify the risks. Information sources times product complexity becomes a geometric increase in the required skills of the underwriters.
The answer to this predicament lies in cognitive technologies that can read, understand, and learn from both structured (text) and non-structured (photos, videos, handwriting) data. Using a computer to read all of the available data, review what actions were taken in the past in similar circumstances, and present alternatives in a matter of seconds is revolutionary to the underwriting process. It solves the time element challenge, the specific expertise talent, and will increase the thoroughness of every transaction processed.
Insurers are investigating today how to use AI to review broader sets of data and more effectively underwrite insurance across multiple lines. The insurers who stay ahead of AI will undoubtedly win market leadership in fast response to applicants and more accurate underwriting, leading to lower claims payments, and higher profits. Progressive exploited analytics in the last decade to build such competitive advantage in the US market. Who will make the investments to do the same today with AI?
Learn more about IBM Insurance solutions here.
 https://en.wikipedia.org/wiki/History_of_insurance, History of Insurance, Wikipedia
 Insurance Underwriter https://www.investopedia.com/terms/i/insurance-underwriter.asp#ixzz59AiU4dL8
 Deloitte US FSI Life Insurance Operations Book of Metrics