Insurance has always been about using data to predict future outcomes. Centuries ago, Lloyd’s of London started by considering basic information such as sea route and vessel type. Since then, however, many structured data sources have emerged that can help evaluate and price risk more accurately. With the emergence of technologies such as cloud computing, IoT, and AI, these sources are only increasing in number and skill at predicting risk. In a matter of years — maybe two, maybe 20 — highly predictive data may be instantly available for every property risk we seek to insure.
As we shift toward this new world of near-perfect data, property insurers may soon find themselves in an era of pre-underwriting, where every potential risk can be computed and priced before a customer ever comes knocking. Carriers can decide how to weigh such risk attributes within the context of their portfolio and then automate underwriting decisions for a large proportion of their book of business.
Pre-underwriting would be a large shift for insurers, altering everything from their profitability to their business models. Imagine, for example, if a carrier could look at the entire state of Texas and request 100,000 home addresses that would meet or exceed a given return threshold? Fully proactive risk selection could be the biggest thing to happen to property underwriting in a century — and it may be here sooner than we think.
From a technological standpoint, the industry is very close to having the fundamental capabilities to institute pre-underwriting for properties in the U.S. Today, cloud computing makes it possible to run an underwriting rules engine across an entire property portfolio. However, insurers must get comfortable with the intricacies of data science that would accompany this shift. For example, insurers may have a lower fidelity understanding of certain details at a per-property level when compared to in-person inspections, but have access to a level of scale, algorithmic consistency, and overall accuracy that would still drive massive underwriting efficiencies, as well as significant book-level improvements in expected loss.
From a data standpoint, carriers interested in pre-underwriting would need to be able to answer the most important questions that are correlated to risk in a highly automated and scalable way. In the homeowner insurance workflow, out of the two or three dozen critical assessment questions, a subset can be filled with tax assessor data. Today, the remaining data points generally come from a homeowner questionnaire or an inspector. However, with new sources of data and artificial intelligence, carriers will be able to access nearly all of this data instantaneously and non-intrusively.
For an external view of properties, carriers can use imagery and AI to better understand overall building condition. At Cape Analytics, we’ve made strong headway in this arena. For an interior view, IoT devices can now detect factors related to safety, security, fire readiness and, soon, perhaps even plumbing and electrical condition. Our digital footprints provide a massive database of demographic information, while weather analytics and AI companies can provide a granular view into property-by-property catastrophe and hazard risk.
Finally, from a regulatory perspective, new risk signals from these alternative sources will need to be filed and accepted as rating variables. Such standards vary from state to state, but some are already leaning forward when it comes to adopting new, non-discriminatory risk signals. As these states test, regulate, and learn from new data, others will follow.
Given all of the above, if pre-underwriting were to become a reality, we can expect to see fundamental changes across the archetypal property insurance organization. It’s interesting to consider how some of these changes may evolve:
Effective pre-underwriting would allow insurers to instantly price policies for every property. The underwriting process would become far more efficient, automatically declining untenable submissions and weeding out false or fraudulent applications that do not concur with the latest data. This would also reduce all the frictional costs that come from uncertainty. As confidence in pre-pricing increases, insurers could downscale or eliminate their post-bind adjustment processes altogether. Pricing would become far more competitive due to these process savings.
This may be a more long-tail extrapolation of what is possible, but pre-underwriting may also provide insurers with the ability to fundamentally reshape their risk portfolio over time. For example, a carrier could proactively target and bind 5% of their book as new policies that are far more profitable than the portfolio average, while simultaneously identifying and either re-pricing, non-renewing, or deploying risk mitigation measures to the bottom 5%, least profitable policies. Over the course of several years, this has a larger potential to reduce aggregate losses than nearly anything else on the horizon. It’s not unthinkable that insurers who adopt pre-underwriting early could improve their combined ratio by 5 percentage points, doubling net profit margins even for a well-performing carrier, while competitors who do not adopt pre-underwriting would face adverse risk selection.
Insurers as front-line risk mitigators
As insurers gain a more perfect understanding of risk, a greater proportion of their efforts may turn towards risk mitigation. Better information allows for the customer conversation to increasingly move from claims to the prevention of loss. This transparency can remove friction from the insurer-customer relationship as insurers can act as a more effective partner, counseling homeowners, for example, on how to reduce their risk throughout the life of a relationship. These conversations can also be bundled with discussions around premiums and credits, and how certain property features contribute to premium. Insurers can focus entire workflows on identifying properties that have risk mitigation opportunities, offering to help them pay for a percentage of feature renovations or replacements, while also offering lower pricing than any other competitor on the market.
Some of this is already happening in health insurance, where insurers may provide price credits for healthy behavior, or even pay for amenities like gym memberships and electric toothbrushes that, if used, lower the chances of a future claim. An even better example exists in personal finance, with the democratization of credit scores. Consumers can now elect to see their credit score, for free, at any time — and companies are competing by providing better resources and education around what causes scores to fall and rise. Entire companies, like CreditKarma, have been built on this premise. Even with this new-found transparency, loan pricing and interest rates vary between providers. A similar dynamic will likely occur in property insurance, even with the existence of pre-underwriting.
Shifts in organizational focus
As straight-through-underwriting becomes the norm, insurers may even be incentivized to shift resources away from the bottom of the insurance workflow — primarily claims — towards the top of the insurance workflow, primarily marketing. When insurers become confident in the risks they are holding, they may choose to spend far more money on customer acquisition in order to take on more good risks. This would lead to far more spend at the front of the funnel, in marketing and underwriting, and a reorganization of the insurer workforce.
Early movers will have their pick of risks to take on — it’s now up to their marketing machine to attract the best available risks, in many cases, pulling the best customers away from competitors. As has already happened in the auto insurance market, customers may move between competing carriers more often as some insurers offload bad risks and others proactively seek good risks. A shift towards marketing also means that customers will be bombarded with advertising, which will also increase their propensity to shop around.
Challenging policy implications
Due to imperfect information, most insurance marketplaces — including property insurance — use risk pooling to distribute risk across various axes of information. In fact, as a result of imperfect information and risk pooling, insurance marketplaces have a secondary impact of redistributing wealth (over-charging some customers, and under-charging others relative to the risk they present). However, once insurers can access near-perfect information, they can be far more granular in how they price policies and which risks they decide to take on.
This increased granularity means that some customers will see higher prices, or may even be dropped by their insurers. Competitive pressure will naturally push insurers to further personalize pricing in order to optimize their business. Theoretically, this would dismantle some long-held risk pools and may have negative societal implications. Rating based on pre-existing conditions is one example of problematic price discrimination particular to the health insurance market.
It is unclear how social policymakers will react to these pressures to break down risk pooling. When pre-underwriting becomes possible, policymakers will likely be involved in answering these challenging questions. Admitted carriers, in particular, will need to think about filed rate constraints and how these technologies can inform rate making. In these cases, state-by-state regulatory discussions will be critical.
As has been the case for centuries, changes in available data will work their way through the insurance and regulatory system and have profound impacts on how the market operates. From our current vantage point, the risks and opportunities presented by pre-underwriting are broad and fundamental. Given the current state of technology and how quickly we seem to be nearing such capabilities, it’s time to start thinking about the potential impacts.
Ryan Kottenstette is CEO of Cape Analytics, , a California company that uses AI and geospatial imagery to provide insurers with instant property intelligence. Look for him on Twitter: @rkottenstette.
These opinions are the author’s own.