Editor’s Note: Collecting and analyzing mass amounts of data for all-size insurance carriers is critical for establishing accurate pricing variable guidelines, which help boost business and ensure a competitive market standing. Today’s carriers rely on granular pricing strategies, which require lots of data, and turn to insurance tech solutions like gradient boosting machines and neural networks to help design risk-based pricing structures. Keeping up in the data arms race can become overwhelming, so many smaller carriers use advisory organizations; however, these third-party vendors can be expensive and often don’t deliver cutting-edge pricing variables.
Data is undoubtedly the basis of designing pricing variables, and insurance carriers must practice due diligence in selecting the best option to deliver claims savings and business growth. Today, all customer data must be protected and any business failing to do so runs the risk of penalties, fines and irreparable reputational damage.
Summary:
Today’s insurance professionals face challenges when pricing policies. Designing pricing variables that take advantage of the newest technologies and data sources can help attract the types of business desired and do a better overall job of pricing.
Many of today’s insurance professionals lack the time, resources, data and technology to create analytical tools within their organization that allow them to more effectively price property risks and compete against larger carriers. Designing new pricing variables, especially pricing variables that take advantage of some of the new technologies and data sources available today, can help reshape a book of business, attract the type of business wanted and do a better overall job of pricing.
But where to start? There are challenges faced when pricing policies. Some small- to medium-sized carriers don’t have the kind of data needed to make the price right, or don’t have the human capital to develop something on their own. If a carrier has a small annual premium volume (say, $500,000 or less) in their book, for instance, it might not be enough to produce the statically credible data needed to create new rating variables.
Pricing granularity
Understanding property characteristics is important for risk assessment. As an illustration, using less granular pricing approaches, it is possible that two homes that are the same age on the same block with identical replacement costs might have the same annual homeowner’s premium. But many characteristics of those two homes that can have a significant impact on risk may be different. Examples include roof coverings, number of rooms, lot size and square footage.
Take a two-story home and a ranch-style home as an example. Through data and research, it has been found that an upstairs bathroom in a two-story home can have a tremendous impact on the risk of non-weather water damage compared to a ranch-style home. If there is an upstairs bathtub, sink or toilet that overflows, that water is going to flow down to the first level, and if it’s not caught quickly, there is going to be extensive damage to the floors, ceilings, and drywall. There is still going to be damage to the ranch-style home, but not to the extent there would be in a two-story home.
The underwriting process becomes clearer with the use of big data and AI.
Another case related to weather is the roof exposure. A large hail event tends to impact the ranch-style home a lot more than the two-story home because the roof size is a lot bigger on the ranch. Those are the kinds of things an insurer needs to be thinking about as they look at their book of business and the exposure, they write for these homes to be priced accurately. The use of big data and AI makes the underwriting process clearer. So, one home could have a 20% surcharge while the other could have a 20% credit instead of having them pricing at the same level.
Accurate risk assessment requires the right detailed pricing variables, which have traditionally been hard to collect. The reality is that historically, carriers had to make coarse pricing and risk assessment decisions, resulting in varying degrees of appropriateness in pricing. Technology and data collection methods are changing dramatically and small- to medium-sized carriers are turning to insurtechs to provide technology solutions.
Big data can help carriers understand the risks
One of the biggest challenges insurers face today is the sheer amount of data available to design new variables. Twenty years ago, when the insurance industry was in the early adoption phase of predictive modeling approaches, the technology and data resources needed to do robust data gathering, processing and analytics did not exist. Most models were based primarily on internal data, when it was available, and it was easier to gather.
Now, insurers use advanced techniques like gradient boosting machines and neural networks. They can cull through and process this massive data to understand what it means and decide what to do with it from a pricing perspective. There’s also an enormous number of data sources available today that can help a carrier better understand hurricane, wildfire, or flood risk, in addition to the very detailed data sources on the characteristics of a home.
The data from connected devices can be invaluable, but a carrier needs to know what to do with it from a pricing standpoint.
The internet of things has also entered the equation. There are all kinds of connected devices—water sensors, thermostats, and burglar alarms—to generate massive data sets that can be a real challenge to deal with. It’s great to have these data sources as they can be invaluable, but a carrier needs to know what to do with the data they produce from a pricing standpoint.
How insurers develop pricing variables
Carriers that are large enough to be able to gather credible amounts of data want to protect their intellectual capital. In addition, they have the tools and human resources to do the analysis, so they will do it on their own.
The key to developing a granular rating structure is significant amounts of data, especially if the analysis is going to be done in a predictive modeling framework. It is also important to know where the data came from and what controls were in place as the data was gathered. Is the data subject to manipulation by a policyholder or an agent? Can they sway data reporting? Does the carrier understand all the characteristics of the data and what it means?
Clients may be analyzing certain data elements and then realize that the data being reported is something different from what they expected. Carriers need to understand what that data is and have enough of it to get credible results. They need to make sure they have the resources, particularly the IT resources, needed to gather the data and conduct the analysis.
They also need to have the experts on staff that can do the modeling, evaluate the predictive power of the data and understand how one variable might interact with other variables in the rating plan. If there is a strong correlation between the variable they are analyzing and something that’s already in their rating plan, maybe it’s not worth it.
Lean on an advisory organization
If a carrier doesn’t have the amount of data or resources they need, there are other options. They may have to rely on an advisory organization that can do the due diligence and make the filings on behalf of the carrier. This approach is pain free from an implementation perspective. The flip side is that these advisory organizations often charge hefty fees for their services. They are also frequently late to the game and tend not to be the most proactive in developing cutting-edge pricing variables. When they do come into the game, they are marketing to hundreds and hundreds of carriers, so any uniqueness or first-mover advantage is lost.
Vendor options
The vendor has done all the heavy lifting around building the variable and doing the testing, but that costs money. What does that mean for the carrier’s ROI? Is the carrier able to do robust testing, understanding the impact on their book of business?
It is important to understand what data sources are out there from different vendors. Carriers need to think hard: Are these data sources going to give them the lift they are looking for? By investing in one or more of these options, is it worth it in terms of savings in claims and if so, will it give them the ability to reshape and grow their book?
This article was written by Klayton Southwood from Insurance Journal and was legally licensed through the Industry Dive Content Marketplace. Please direct all licensing questions to legal@industrydive.com.
Ensure the security of your data with Windstream Enterprise cloud-optimized Managed Network Security (MNS).
