Sussing out Risk

When it comes to uncovering information, identifying patterns and revealing truths, not everyone can be Detective William Murdoch – innovation and prescience personified – the fictional sleuth of Murdoch Mysteries whose wits unfailingly win the day. Far too often, in insurance, as in whodunnits, conventional methods are the norm. But who wants to play the role of the elegant detective’s plodding counterpart, Inspector Brackenreid?

Detective work around the turn of the 20th century and commercial property insurance of the early 21st century are not as far afield as might be imagined – for both, it is all about the data and analytics. An explosion of third-party providers of data and analytics has many commercial insurers reassessing their traditional methods for capturing information about potential risks.

GATHERING CLUES

In general, insurers have an obligation to their owners, distributors and clients to be both thorough and innovative in their sales and underwriting processes. The more complete the information about a property, the more competitive and sustainable an underwriter’s pricing will likely be.

Consider this example of insurance detective work. A hypothetical insurer is evaluating a commercial property for coverage, and the underwriter knows it is a 20-unit motel, two storeys tall, of masonry construction and located in Ontario. In this imaginary case, the insurer previously covered the property from 2008 to 2010, with one paid claim on file.

That hypothetical profile sounds pretty straightforward, although there can often be inaccurate or missing information, if not outright “unknowns.” For example, what if from 2010 to the present, the insurer that covered the property received a fire claim? Also, there were two more fires – one in a dumpster at the rear of the building and another in the laundry room caused by an overheated clothes dryer.

Neither incident resulted in a claim, and the traditional process for developing the underwriter’s information would not have necessarily disclosed that fact. Fire and gas incidents may suggest the presence of risks and a lack of precautions (previously unknown to the insurer) that warrant urgent examination.

Across the insurance industry, data about fire reports often remains a big question mark. The National Fire Incident Reporting System (NFIRS) collects incident reports in the United States from the majority of the country’s more than 50,000 fire jurisdictions. As a result, the NFIRS database represents about 75% of estimated fire incidents – a substantial majority, but still incomplete.

Unfortunately, Canada’s more than 3,700 fire departments have served without a central reporting clearinghouse for more than a decade, making an underwriter’s decision whether or not to write the risk – and how to price the policy – even more challenging.

Other unknowns often include the property’s weather history. What is the likelihood of prior hail damage that has not been reported? How frequent are wildfires or lightning storms? What are the chances of additional peril events in the future? Predicting those would likely be a tall order even for Detective Murdoch, let alone any underwriter.

TRACKING CONTEXT

Granular and up-to-date analytics on a range of perils – wind, lightning, hail and wildfires – are a key requirement for assessment of weather-related risks to a property.

Throughout Canada and the U.S., there is already a detailed network of ground-based radars (for wind and hail), lightning detection sensors, satellite and airborne remote sensing (for wildfire mapping, terrain and road network analysis), weather and climate models, as well as ground observation stations (for further data-analytic validation). These data assets cover a wide spectrum of granularity.

Working together, these sensors collect data at an astounding velocity. Radar networks observe fast-moving storms about every five to 10 minutes, in a system that includes the roughly 31 individual radars covering the most populated regions of Canada. A global network of about 700 lightning sensors records more than 99% of all cloud-to-ground strikes. That information is further augmented by observations from more than 1,000 weather stations in Canada.

Peril data developed by Verisk Climate includes a trend indication for each of the events, historical occurrences down to the street level of the property, and date of the most recent event – all of which help provide insight into the probability that a damage-producing incident affected the property in the past.

Every detective story has a dénouement, and insurance stories are no different. For many underwriters, the question of whether a property fits their appetites, or comports with their guidelines, can often be answered with several pieces of evidence:

• likelihood the property has pre-existing damage or a history of incidents;

• need to perform a property inspection; and

• the property’s risk level compared with its peers.

Most insurers need new information on perils and incidents to better understand risks, align those risks with their underwriting guidelines, and prioritize inspections for properties with a history of incidents that did not result in a corresponding claim. But raw, unexamined data can often only go so far – there needs to be careful analysis to translate that information into something useful and actionable for an underwriter. Scoring the likelihood of future weather-related events can add an entirely new dimension to the review.

Weather data and analytics of the type available in the U.S. could be a similar game-changer for many Canadian insurers. Although record-breaking hailstorms seem to be de rigueur for Calgary and its environs, it is not only the historic storms that are resulting in claims. A steady stream of smaller hailstorms over the course of time can have the same effect as one large event.

Four storms of low to moderate damage may not lead to a filed claim, but the damage over time may, ultimately, result in one. The insured could file the claim after the fifth storm has inflicted similar, but compounded, damage.

Many insurers are increasingly recognizing the unprecedented granularity and accuracy of these assets to empower more effective underwriting and claims management. Underwriters can often ill afford to be the insurer for that last storm – the storm that leads to a claim – when they could have known about the previous four.

As Detective Murdoch demonstrates time and again, it can pay to do some careful legwork, review the clues and follow new insights to success.