By Chris Early
The use of analytics has become table stakes in the insurance industry in general but especially for personal lines auto. Since the mid-1990s, the number of personal lines auto insurers has decreased by a third, and direct written premium for the personal lines auto insurance industry grew steadily until reaching a plateau in 2004. In the meantime, the number of insurable vehicles continued to grow year over year, which has resulted in lower premiums per vehicle in more recent years (see Figure 1).
This is not a coincidence.
In part, the lower premium per vehicle is due to safer vehicles and an aging population that tends to have lower loss costs. But undoubtedly, other contributors to lower premium per vehicle include increased pricing segmentation, more accurate risk assessment, and less average pricing by the industry — all a result of the use of more sophisticated analytics.
Before the mid-1990s, most insurers used the same methodology on the same risk factors. This usually consisted of a one-way loss ratio relativity analysis on a particular factor to determine if some defined segments within the factor were performing better or worse than the factor average. If the data was fully credible, the insurer made the changes to the factor as indicated by the one-way analysis. However, if the data was not fully credible, the insurer would move toward the indicated changes based on the credibility percentage (the square root of the number of claims/1,082) and then assume current factor selects were accurate for the remainder (the complement of credibility, or 1 minus the credibility percentage). Rarely was consideration given to how interactions between factors could affect the accuracy of the results, and even less frequently were they accounted for in pricing.
But a pricing revolution started in the mid-1990s that unleashed the power of analytics and demonstrated its capability to level the competitive playing field. New risk assessment factors (such as credit-based insurance scores) and statistical modeling (such as Bailey’s minimum bias) came into vogue for personal lines auto pricing — which changed the industry forever.
The evolution in pricing accuracy has continued to accelerate beyond the first decade of the new millennium because of more available data sources, rapidly advancing technology, and even more sophisticated analytics employed across more lines of business. Insurers that embraced this change have thrived and gained market share, while insurers that did not act or did not act as quickly have struggled to maintain market share, have been acquired, or are out of business entirely.
As noted above, the shakeout among insurers decreased the number of auto insurance carriers in the marketplace, which has led to greater market concentration among fewer insurers. As Figure 2 shows, the top ten auto insurers accounted for 56 percent of the market in 1995 but now account for nearly 70 percent. Carrier numbers 3 through 10 have grown from 23 percent of the market to 39 percent. Not surprisingly, and as many industry observers can attest, most of the insurers in positions 3 to 10 were early adopters of sophisticated pricing analytics. But they didn’t limit their use of analytics exclusively to pricing; in many cases, they also used sophisticated analytics in marketing, underwriting, and claims.
Cumulative Market Share by Insurer Ranking
In addition to the use of analytics being a “must have now,” insurers also recognize that analytics can be a game changer and can quickly level the competitive playing field. If you’re one of the top ten market leaders, you need to be aware of what’s going on in the world of analytics, or you may wake up one day and wonder what happened to your book of business.
If you’re a middle-tier insurer or a smaller insurer, the implementation of the right analytics solution can help you catch up to the larger insurers more quickly, compete more effectively, or simply survive in the world of giants. Thus, all insurers should routinely assess their current analytic situation, stay abreast of what is happening in the marketplace, and determine their needs regarding analytics. Then decide how to proceed.
Generally speaking, insurers have six options concerning how they’ll proceed with analytics (see Figure 3). While some insurers might use only one option, most insurers choose to use more.
|1. Use In-House Resources||• Readily available data and staff
• Data is specific to insurer’s book
|• Possible data limitation for certain segments
• Don’t know what you don’t have
• Large, ongoing investment
|2. Build In-House||•Controlled environment with improved tuning to the current book
•Ability to expand or reduce the team to suit the business need
|• Limited data and book size
• Large, ongoing investment
• Slow speed to market
|3. Engage a Consultant||• Third-party perspective on industry landscape and competitor benchmarks
• Flexibility to engage and disengage as needed
|• One RFP at a time with high fee structure
• Internal actuaries not optimized
• New RFP and new search
|4. Copy Another Carrier||• Possibility of “riding the wave” of a successful competitor
• Focus on other competitive differentiators (e.g., marketing)
|• Mismatched risk profiles
• Update to book delayed until filings are made public
|5. Buy from a Vendor||• Improved risk model from large data set
• Timely updates and recalibrations
• Expand focus beyond loss cost analytics
|•Potential concern around viability of longer term partnership
• Lack of updates to maintain model’s superiority
|6. Do Nothing||• Continue to reap current market advantages
• Limited incremental investment over the short-term horizon
|• At risk for adverse selection
• Suboptimal long-term plan for sustainability
• Suboptimal marketing spend
Regardless of which option(s) insurers choose, they need to be aware of the resources available to them and what their competitors in the marketplace are doing. With the rapid advancement of technology, the increased sophistication of analytics, an evergrowing number of data sources, and the untold number of terabytes of information that these sources have available, developing and implementing the best analytics solutions will not be a simple task, but it’s a “must have now.”
Chris Early is assistant vice president of analytic product management and customer strategy at ISO.