So What is Underwriting Anyway?
Apart from a certain subclass inside the insurance industry, not many people know what an insurance underwriter does. This applied to me when I joined AIG as an underwriter trainee in the mid-90’s. I gave up after a couple of attempts to explain to my parents exactly what I would be doing. My experience has been that Insurtech is not immune from this either, particularly since many members of the Insurtech community come from outside the system.
Underwriting can be confused with actuarial work. Actuaries set the rates or prices that the company must charge. And they set reserves on the balance sheet, which are the amounts put aside to pay for future claims. Actuaries crunch data with sophisticated methods of statistical analysis, aka “math”, to isolate loss trends and correlations. This information then informs the underwriting decisions an insurance company makes.
OK, so what does the underwriter do?
Underwriters apply the abstract work of the actuary to the real-world case of an insurance applicant. The actuary may be able to say, with statistical confidence, that a white-linen restaurant in Cleveland, Ohio should pay $1.85 premium per $1000 in business revenues, but the data may not exist to differentiate between risks that offer valet parking and those that don’t (just to create an example). It becomes the underwriter’s job to apply the actuary’s rates to the quote-seeking applicant, who may present variance within the subset of the statistical model.
An example based on profile:
Two fleet auto risks have the same number and type of vehicles, the same geographical use area, the same type of business and loss history. One of them, however, just lost its senior risk management staff; while the other just completed an industry-leading driver training course and an upgrade to its fleet management software. When the broker asks for a price discount, which of these two risks is more deserving, and for how much? The underwriter makes that kind of decision.
An example based on exposure:
A textile manufacturer converts its production lines to produce Personal Protective Equipment in support of the healthcare industry and its Covid-19 challenges. A CGL policy covers the products and completed operations liability for the business. On renewal, the underwriter must decide if an exclusion would be appropriate for any claims related to pandemic/viral transmission, or if that risk should be accepted with an increased premium, or if the account should be nonrenewed. There is no “correct” answer, as much depends on the company’s risk tolerance and marketplace strategy. What would you do, presuming your goal was to produce an underwriting profit?
Underwriters will engage in debate about whether the tasks they perform are an art or a science. Or to rephrase, can the selection of risks be reduced to purely data-driven decisions or is there room in the process for the underwriter’s intuitive knowledge, and where does that intuition come from? This debate often ends up being reduced to semantics, particularly as you get later into the evening. You will hear experts and sages contradict each other with “… always trust your instincts”, or “… never go against the data”.
My thoughts are that underwriting is like an artisan craft developed over time with experience and the use of good judgment. Every underwriter with a decade or more of experience has made some good calls and then been burned by some of the bad (and costly) ones. Both the good and bad experiences are beneficial, so long as good insight and increased awareness result.
Put this into the context of the Insurtech companies in the underwriting game, often using new models with limited actuarial support. The underwriting results from the initial players haven’t been great, but of course, this is all new, so no judgments offered. “The jury is out”, as they say. An insurance company can operate at a loss for as long as its capital providers tolerate the negative ROE.