P&C Has a Problem With Classification

SIC was set up in the 1930s, for an industrial economy. Try looking up the code for a web developer. Just try.

sixthings
The commercial property and casualty (P&C) insurance industry has a classification problem. The overlapping use of standard industrial codes (SIC), Insurance Services Office (ISO), National Council on Compensation Insurance (NCCI), North American Industry Classification System (NAICS) and a dozen or so other carrier-imagined coding systems are creating a growing number of problems for insurance companies and agents alike. Ambiguity, misuse, line of business (LOB) specificity, misunderstanding and straight-up miscategorization leads to missed sales opportunities, higher underwriting costs and unexpected exposure to risk down the road. It’s time for the industry to examine risk categorization and take a fresh shot at solving the classification problem. The problem is, many of the fundamental issues stem from limitations of the underlying classification systems themselves. Dude, SIC Standard Industrial Classification (SIC) is like the mainframe computer of classification systems. Wildly out of sync with the modern world, it somehow is still extraordinarily prevalent in the wonderful world of insurance. SIC was first established in the 1930s as a way for government agencies to speak the same language with one another. Obviously, American businesses of the '30s, '40s, '50s and '60s looked very different than the business of today. SIC was developed mostly as a taxonomy for an industrial economy. Attempt to look up the SIC for web developers, for example, and you won’t find it. Many other modern businesses are also missing, because the SIC system itself was retired in 1997. SIC includes broad classifications and generalities, which are largely too unspecific to be of much use in classifying risk. Consider a fairly common code: 5812: Eating Places. Naturally, one would think immediately of “restaurants,” right? Unfortunately, that code also encapsulates industrial feeding, dinner theaters and sports arena concessions. While many insurers would likely write the GL on a restaurant down the street, few would be inclined to touch industrial feeding (whatever that is). What’s good about SIC is that it’s insurance agnostic and has applicability to all lines of business. What’s bad about SIC is that it’s too broad and, perhaps even more of a deal breaker, it stopped being updated two decades ago. See also: Future of Securities Class Actions   The Last DJ There’s a line in a song by Tom Petty & the Heartbreakers that wonders, “the boys upstairs want to see how much you'll pay for what you used to get for free.” This comes to mind anytime anyone advocates for a proprietary classification system. Setting aside the often-high licensing cost, these schemes are also almost always hardwired to the insurance industry, often because they have roots in rating. There are at least two significant consequences to this hardwiring. First, it creates a user experience challenge. This manifests itself to the policyholder, be it on an insurance form or a website. For an industry outsider, categorizing a business using an insurance code is just plain hard. This user experience deficiency also extends to the agent or CSR who must first be trained in the codes themselves before being able to begin to apply them. Second, and of increasing consequence, these values are not found in publicly available data. As insurers and intermediaries continue to integrate with any and all available public data to improve underwriting and reduce sales friction, proprietary, industry-specific classification is always something that will have to be derived after the fact, creating an exposure for error. With the continued use of proprietary classification schemes, industry participants are paying a bundle to speak a language that’s foreign to the rest of the economy. LOB-Specific Systems Workers’ compensation (workers’ comp) is perhaps the best example of a LOB with a classification system developed for pricing. Workers’ comp codes are exceptional for pricing workers’ comp. These codes do the job perfectly, but a coding system designed to attribute premium to payroll classes is not the same thing as a classification system to categorize the aggregate risk associated with the operation of the business. As an example, any insurer that has ever written workers’ comp will know that two of the top “classes” are clerical and outside sales. But what does that business with those exposures actually do? Is a business with clerical exposure an office full of desk workers, or is it a couple of managers at an industrial chemicals factory? The problem with workers’ comp codes, in particular, is that they describe what specific employees at the business do, not what the business itself does in the big picture. Expressing risk appetite is challenging, if not impossible, using payroll codes as a basis, and it’s difficult to convert these codes into eligibility for other LOBs for cross-selling other lines. A higher-level description of what the business actually does is required for portfolio underwriting — payroll codes are not sufficient. The case for NAICS NAICS is provided for free by the U.S. Census Bureau. NAICS classification provides very detailed descriptions of what it is that the business actually does, and, thusly, what exposures come with its operation. NAICS replaced SIC in the late ‘90s. In terms of detail, SIC codes are like on old tube TV, and NAICS is like a 4K flat panel. This high-definition classification has been updated to reflect the service economy and has a much lower level of detail describing exactly what a business does. The eating places conundrum discussed earlier in the context of SIC is a great example. With NAICS, it’s actually possible to code the restaurant down the street differently than a snack vendor at a stadium. As an insurer responsible for underwriting such risks, a high-definition description of the business operations is tremendously useful for pricing and underwriting. Clearly, not all eating places have the same exposures, so NAICS is a fantastic way for commercial insurers to express appetite. (Cafes, yes. Dinner theaters, no.) NAICS is also updated on a five-year rotation, most recently in 2017. Each revision builds on the last, incorporating new and emerging business types. Furthering the case for NAICS, the U.S. government has incorporated it as the go-to system for business classification, meaning that publicly-available data sources can and do provide these codes. See also: Chatbots and the Future of Interaction   A Call for Class Action Adoption of NAICS as as a standard for commercial P&C risk classification benefits the industry in all phases of the policy life-cycle — from reducing friction in how products are distributed; to reducing exposure to mis-categorized risk; and to enabling portfolio underwriting, all while supporting the new businesses being created in the modern economy. It’s long past time we all got on the same page.

Michael Albert

Profile picture for user MichaelAlbert

Michael Albert

Michael Albert is the co-founder of <a href="https://www.askkodiak.com/#/home">Ask Kodiak</a>, an appetite as a service platform for commercial P&amp;C. He has nearly two decades in the P&amp;C insurance industry, serving in senior technology leadership roles in product and strategy.

MORE FROM THIS AUTHOR

Read More