Underwriting in the Evolving Human and Algorithmic Landscape

Stephen Sills, former Chairman, and CEO of CapSpecialty, discusses his journey in the US insurance industry, and the challenges and opportunities of underwriting in the evolving human resources and algorithmic landscape.

Interviewer notes:

I had the honor of meeting Stephen a few years ago when he was Chairman and CEO of CapSpecialty. Although we rarely had an opportunity to meet, his vision was one of the guiding lights that led me into commercial insurance. Now that Stephen has retired from CapSpecialty, he had the time to discuss his journey in the US insurance industry and the challenges and opportunities of underwriting in the evolving human resources and algorithmic landscape. His reflections are inspiring.

David Schapiro (DS): Could you please tell us how you got into the insurance business?

Stephen Sills (SS): Like most people, I didn’t grow up longing to get into the insurance business. After hitchhiking most of the way around the world, I went looking for a job. I had friends who were brokers specializing in insurance for fine art and jewelry.

Insurance sounded interesting, so I went over to the College of Insurance (one of my friends had graduated from there) and asked if anyone was hiring. They gave me the name of Stewart, Smith Aviation – wholesale brokers and MGA underwriters for unusual aviation risks, which frequently required London market expertise and capacity.

I did that for two years, until I was asked by a senior executive at Stewart, Smith – Joseph Barrett – if I would be interested in working on Directors and Officers Liability (D&O) insurance. Our company was underwriting on behalf of CNA at the time.

Joe was a mentor extraordinaire. He spent hours every day teaching me policy forms, securities law, and underwriting skills. Not only was I learning a lot, I was being paid for something I loved doing.

DS: Once you got into insurance, how did the journey continue?

SS: After more than 10 years in the industry, I had the opportunity to start Executive Risk (NYSE:ER), an insurance company that initially wrote only D&O insurance. We grew to be a multi-line specialty insurance carrier with over $600 million in premiums. After over 10 years at ER, I retired subsequent to its purchase by Chubb.

After a few years of retirement, I had the opportunity to start Darwin (NYSE:DR) with the support of Weston Hicks of Alleghany (NYSE:Y). Darwin became a full-service specialty liability insurance company with in-house expertise in underwriting, claims management, risk management, and information technology.

DR was subsequently purchased by Allied World, and I retired for a few more years before joining Weston Hicks for the turnaround of an Alleghany subsidiary named Capitol Transamerica in 2013. I became Chairman and CEO, and we rebranded the company as “CapSpecialty”, focusing on Property and Casualty, Professional Liability, Surety, and Fidelity products for small and midsized risks.
On July 1st, 2019, I retired.

DS: Looking back at this journey, are there some insights about the human resources landscape that you could share with us?

SS: I believe the human resources landscape has changed substantially in the past 40 years. There have been many drivers of this change.

In the 1980s, insurance companies operated management training programs that generally combined on-the-job training with a substantial amount of classroom instruction. Companies like the Aetna Casualty and Surety Company built a large residential conferencing center in Hartford solely for educating their trainees and executives.

Some of these homegrown underwriters left to go to brokers, but because of good relationships and knowing what their old employers wanted, they frequently placed business back where they came from. To fill the holes, there was generally a solid bench of talent. The training programs filled those slots. There were not a lot of insurance startups for underwriters to jump from one risk-taking entity to another.

Around this time, companies became a lot more cost conscious. Expenses needed to be cut. Curtailing, and ultimately eliminating, talent development was a short-term fix that would go on to create a long-term problem.

DS: Have you seen this long-term problem come into existence during your journey?

SS: Thirty to forty years ago, insurance was viewed as an arcane, backwater industry. Things got done a particular way because it had always been done that way. New risk-taking capital was attracted to the business only if there was a need to remedy a lack of capacity. Capital to support innovative risk taking or processing was rare.

In the past 20 years, the industry has substantially changed because of terrorist and hurricane disasters. Not only was the capital of the industry impaired, but while standard markets spent several years recovering, new risk-taking capital flooded in to support existing risk takers and also fund new, fleeter risk-taking entities.

Where did the talent come from? It was traded among the existing carriers to the point where we read every day about whole teams of people leaving one carrier for another.

DS: How have you seen the underwriting process and profession change over the years?

SS: Over the last 10 years, I have seen companies focus ever-greater energy towards reducing expenses. One of the most important areas to focus on has been Information Technology (IT).
While once only used for record keeping, companies began to see that IT could be used in the underwriting process to ease the paper burden of underwriters and make in-house data available for decision making. We now see the intersection of technology and underwriting, such that the underwriter is not just aided by technology, but rather replaced by it.

For a basic risk, a company in possession of thousands of similar risk profiles need not train an underwriter to evaluate the exposure. The underwriting system can give the required terms and conditions without management’s fear of the underwriter making an exception to a rule or perhaps making a decision based on a personal relationship rather than the hard facts.

DS: What kind of impact did technology have on the industry during your journey?

SS: We have seen personal lines insurers successfully avail themselves of cutting-edge technology for underwriting and distribution of their products. Pioneered on a large scale by companies like Progressive and GEICO, personal lines insurers successfully utilized predictive analytics to accurately select what risks they wanted and those that they wished to avoid. The machines made the decisions based on massive amounts of data.

The days of only looking at age, gender, moving violations, type of car, and location are long gone. Establishing a price based upon an application completed by the driver’s agent did not get optimal results. In years past, an underwriter would intuitively know that the number of hours you drive and how you drive (even if you are not stopped by the police) affect the risk, but they had no way of accurately gathering that data or, if they did have it, of effectively utilizing it.

With the advent of telematics and sophisticated machine learning, insurers can more accurately price to the risk and not the opportunity. With data analytics and the internet, these companies can effectively reach consumers directly, without an agent or broker. The consumer can be comfortable in their decision to purchase without an intermediary because state insurance regulators mandated a relatively homogenous product offering.

DS: How did this process affect the insurance companies’ brand and industry consolidation along the value chain?

SS: Reaching those consumers and building their trust was a massive undertaking for the insurers. Many billions of dollars have been spent getting consumers to trust their insurance company without an intermediary. Having good technology was not enough. Brands needed to be established and continually reinforced by fast, fair claim handling.

These focused, cutting-edge companies were aggressively building their businesses at a time when less-fleet-of-foot companies realized that their cost of writing small policies and their need to build new technology were a drag on earnings. Some made the transition; some exited the business. Some larger companies that were agent focused invested in technology and call centers to take the processing off the hands of their agents who didn’t want to bother with small premiums. Many agents, failing to have enough of a “value add” to justify their commissions or a large enough size (or a plan for generational succession) to justify investing in the technology to integrate with the cutting-edge companies, opted to sell to consolidator agents.

If you were a small premium/volume agent, you had to sell to a larger agent. If you were a small standard insurance company, you couldn’t afford to build the technology needed, so you either sold personal lines books of business or your whole company.

DS: Has this evolved differently in commercial lines compared to personal lines?

SS: The handling of small commercial business is at a major inflection point. To date, very little of the above exists in commercial insurance. Technology has been used to reduce paper and streamline processing, but little has been done by risk-assuming companies to evaluate subjective risk.

Current technology can automatically calculate a property’s distance from the coast for coastal wind exposure purposes, but if the risk were to fry food on the premises, the answer to the question, “How often do you have the fryer exhaust cleaned of accumulated grease?” cannot be known with certainty. To date, there are no third-party data providers that make that information automatically available.

Even less has been done in the Commercial General Liability (CGL) space. There is very little homogeneity with commercial risks. A restaurant is not just “a restaurant” for underwriting purposes. What percentage of their sales are from alcohol? How late are they open? Is there entertainment? What type of entertainment? With the absence of that type of data being provided automatically by reliable data providers, the underwriting of commercial risks will have to continue to rely on manually provided data from agents/brokers/insureds.

DS: How does this inflection point influence commercial underwriting?

SS: The current state of commercial underwriting is in flux. For many years, insurance companies relied on agents and brokers to provide them with subjective and objective data as well as underwrite risks for them. They were the “boots on the ground” and knew their local risks. Many of those local brokers/agents have now been rolled up into national organizations who have centralized the underwriting process away from the local exposure. In the absence of that local knowledge, underwriting quality has gone down, while expenses have frequently gone up.

This environment has seen the offering of online shopping by digital intermediaries but with little uptake. Expenses have merely been reallocated, and the buying experience has been less than satisfactory. A commercial shopping site that accesses five insurance companies forces the applicant to fill out a questionnaire that includes all of the questions from the least efficient risk taker. And, those least efficient risk takers have little incentive to support a shopping site with a more streamlined process than that which they provide their most faithful agents.

Just as consumers have become more comfortable, so too have agents. Around 2005 at Darwin, we rolled out a product we called iBind. It was a program that allowed the agent to complete the application for small private D&O risks online or send a link to their client to have them complete the form online. A bindable quote could be secured, and the potential insured only needed to sign it to secure coverage. The agents were very uncomfortable with this process. They didn’t want to do the form online, and they didn’t want their clients communicating directly with the company.

Fast forward to 2018, when CapSpecialty created DragonX, an online system for agents to rate/quote/bind excess casualty coverage in as little as two minutes. The adoption rate for that product was as fast as I had ever seen for a new product.

DS: Does the level of automation vary across different policy types and premiums?

SS: While I’m suggesting a pessimistic short-term view of CGL insurers becoming as automated as the most efficient personal lines insurers, progress is being made on the underwriting front. Less progress is being made in the direct-to-consumer space for anything other than the most basic/simple risks. One of the main reasons for this is the complexity of the basic policy form and the myriad of possible endorsements offered depending on the unique nature of a commercial risk. Licensed agents and brokers are experts in advising the type and scope of endorsements needed for protection. For an additional cost of 10–20% on a premium that is frequently less than $3,000, insureds have opted to pay more for the peace of mind.

Stephen Sills – Bio:

Stephen Sills is the former Chairman and CEO of CapSpecialty and PRMS, having brought over 30 years of insurance industry experience with a record of proven success and commitment to customers. He founded Darwin Professional Underwriters, Inc. in 2003 and served as its Chairman, President and Chief Executive Officer until 2008. Prior to this, he founded Executive Risk Inc., where he served as Chief Underwriting Officer and Chief Executive Officer. Mr. Sills previously served as a member of the board of the Connecticut Children’s Medical Center and as president of the Professional Liability Underwriting Society (PLUS); and serves on the board of MCC Theater Company in New York City.


Check out our next CEO Series interview with Alan Bauer, retired Progressive Group President and insurance visionary.

Related Articles

We use cookies to provide the services and features offered on our website, and to improve our user experience. Learn more