While the availability of data continues to increase, it is also crucial that carriers translate this growing wealth of information into meaningful underwriting action.
By Mark Schauss, Managing Director, Binding Authority
For property and casualty insurers in the small business arena, the pricing process is rapidly evolving. The burgeoning wealth of readily accessible data on smaller companies—combined with carriers’ growing data analysis capabilities—means that insurers should increasingly be able to offer pricing that more accurately reflects the true profile of each business. Ultimately, however, there is a remaining challenge: how carriers can most effectively implement underwriting strategies that reflect these new capabilities.
To more fully understand pricing’s evolution, it is worth recalling that its history is founded on the simple concept of the law of large numbers. In essence, this says that if you flip a coin a few times, you may see a string of tails and think “tails never fails.” However, flip that same coin 1,000 times and you’ll confirm that when someone says something is “a coinflip,” it does indeed translate to almost exactly a 50 percent chance of it happening. Accordingly, as the number of homogenous scenarios increases, the more accurate we can be in predicting future outcomes, including whether an insured is going to have a loss, and how much that loss will end up costing.
“Pricing for small businesses has historically been hampered by a lack of relevant data on their loss history.”
However, when assessing small business risks, establishing the right categories of “large numbers” can be trickier than you’d think. While there certainly are small businesses that fit nicely into homogeneous groupings—a typical drycleaner, a convenience store, or a corner pub—many small businesses exist to solve a special problem or need. Compare this to personal auto or homeowners insurance, where the dominant use of one’s personal auto or home is largely the same for the vast majority. In a way, many small businesses are like unique snowflakes. This niche characteristic has historically made it cumbersome to match pricing to small businesses’ profiles, typically resulting in a broad-brush pricing approach from carriers.
In addition to diversity of operations, pricing for small businesses has historically been hampered by a lack of relevant data on their loss history. Often you will hear insurance professionals say that the best predictor of future loss experience is past loss experience. This is true when there is a meaningful loss history to examine. On average, only about one to three percent of small businesses have an insurance claim in any given year. By contrast, larger operations tend to have a richer history of past loss experience, which serves as a critical building block for pricing those risks.
Historical underwriting processes for small businesses also impede accurate pricing. Insurers have traditionally relied heavily on the operations of the insured to govern the rating process. This practice dates from a time in the past when every underwriter had a massive three-ring manual on their desk. Historically, it made perfect sense. The insured’s operation or class of business code was then the easiest starting point for articulating underwriting guidelines and a pricing approach.
“Our own internal analysis has shown that the category of operation in certain circumstances has little to no predictive power as to whether there will be a loss.”
In fact, in many scenarios today, small business insurance rates are still only two dimensional: class code plus territory equals rate. And given that some territories cover an entire state, it is very possible that if two companies are in the same business, and located in the same state, they could be assigned the exact same rate, regardless of their other dimensions of risk.
This approach has major limitations, given the wealth of other factors that can help predict future losses. Our own internal analysis has shown that the category of operation in certain circumstances has little to no predictive power as to whether there will be a loss. In fact, in certain scenarios, something as simple as real estate pricing trends in the surrounding area has actually proven to be a better predictor of future losses than the insured’s industry category alone. So while class codes continue to have important non-pricing roles related to reporting and common language, they may no longer be the best starting point for pricing.
Fortunately, a more precise approach is now possible through multivariate digital rating, which can provide both better rates for qualifying insureds and better risk selection for insurers. For example, if Carrier A develops a rate using only class code and territory, while Carrier B develops a rate using class code, territory, and credit score, Carrier B should be in a better position to more advantageously rate businesses with superior credit scores. Meanwhile, Carrier A will be more likely to continue offering similar rates to businesses with good and bad credit scores alike, and will therefore only land the companies with bad ones.
“When assessing small business risks, establishing the right categories of ‘large numbers’ can be trickier than you’d think.”
While the availability of data continues to increase, it is also crucial that carriers translate this growing wealth of information into meaningful underwriting action. For example, technology today may be able to show you a satellite image of a building in California that instantaneously maps its distance to the nearest brush—a consideration when weighing its exposure to wildfire. While this mapping tool may be a fantastic underwriting resource, quantification and guidance must accompany the capability for it to be useful. In other words, what does the number of feet between the structure and the nearest brush actually mean? Should a close distance result in an added pricing factor of 5 percent or 20 percent, or maybe a declination altogether?
To quantify the impact to loss costs of these new data sources often requires generating credibility through “large numbers” of known results, which typically means past results. In this example, retrospectively mapping images onto past known policies may not even be possible; so how best to put these images into action is easier said than done.
Ultimately, carriers need to become experts at translating this bounty of new data into assignable and quantifiable loss costs. Successfully integrating these new data sources into an effective underwriting strategy continues to require the establishment of credibility through the law of large numbers. Carriers who are able to adapt may successfully minimize adverse selection within their portfolios, allowing them to provide qualified insureds with a more efficient and cost-effective insurance purchase.