Sophisticated Analysis

As insurance companies become increasingly sophisticated in their ability to calculate premium rates that more accurately and precisely reflect the risks they are underwriting, it will be important for insurers to educate brokers and consumers on the new techniques they are using.

The Economical Mutual Insurance Company sat down with Canadian Underwriter in March 2010 to discuss the importance of educating brokers and consumers about increasingly individualized – and more accurate – methods for pricing risk.

The issue is top of mind for The Economical, which has reworked its existing auto insurance product, updating the way it is designed and priced. The updated auto product will be launched May 15, pending regulatory approval.
Key to the product update is an increasingly common, ‘multivariate’ approach to pricing risk. The approach essentially moves away from a more generalized, class-based approach to writing risk, focusing instead on the interplay between risk characteristics, thus allowing insurers to derive premium on a more individualized basis.  

What is ‘Multivariate Analysis?’

To some, it may seem like a person has to hold an advanced degree in Mathematics and Statistical Analysis to understand how a person’s premium is derived. And referring to terms like ‘multivariate analysis’ probably doesn’t help to dispel that notion. Simply put, the technique allows insurance underwriters and actuaries to analyze millions of different combinations of ‘variables’ – which in this context would be the consumer’s risk characteristics (age, years licensed, age of the car, etc.) – to come up with a precise, accurate premium for that particular consumer.      

Brokers have expressed concern about transparency related to the rapidly evolving sophistication of the analysis. This concern was raised at an insurance company CEO plenary discussion at the Insurance Brokers Association of Ontario (IBAO) 90th Annual Convention held in Niagara Falls in October 2010.

The brokers’ concerns might be summarized as follows. A broker sends his or her client’s risk characteristics to an insurance carrier. These variables are then input into the carrier’s statistical package, or ‘black box,’ which spits out a premium number. How is the process of obtaining this final premium amount then explained to the consumer?

Or to put it another way: if the underwriting technique for analysis gets too complex, if brokers don’t understand what’s in the complex rating structure, then how will they be able to explain the premium price to a client?

The transparency issue is further complicated by competition between carriers. Private insurance companies have a legitimate business interest in keeping the exact formulae used to determine premium prices confidential – especially if that confers an advantage over competitors.

There is a place where all of the parties can meet, according to The Economical, and it is important to explain all that can be explained to brokers and consumers. Multivariate analysis is not necessarily “new” – major carriers are all using it, and The Economical has been using the method for seven or eight years already – but as the technique is increasingly employed and refined, brokers and companies that cannot keep up will be at a competitive disadvantage.

“It’s been in the industry for awhile and a couple of our leading competitors are quite good in this space, so we need to keep up, otherwise our brokers are at a competitive disadvantage,” says Katherine Mabe, president and CEO of The Economical. “We need to make sure they have the most competitive pricing capabilities on the street. We want our brokers to have the very best tools. If we were to say, ‘We aren’t going to go this way because it’s difficult to understand,’ then we leave [brokers] at a disadvantage. So what we have to do is give brokers as much education as we can to explain generally how it works.”

Increased Sophistication

This might be easier said than done. The techniques employed can refine pricing to a dizzying degree. The ‘Old World’ of looking at variables featured somewhat clunky, cumbersome ‘classes’ of people. A consumer might have seen his or her premium rise or fall simply because of the claims experience of his or her class. For example, a young man’s premium might go down simply as a result of a 25th birthday, which would take them out of the riskier 18-24 category of drivers. It’s not that the customer’s driving got better, it was simply that the class of drivers to which he belonged had a better claims experience.

Multivariate analysis allows underwriters to look at the different effects of multiple factors combined together. And it allows underwriters to account for when the importance of one variable may be amplified, exaggerated or rendered insignificant when analyzed in combination with others. “So it literally becomes millions of combinations of price points instead of 20,” as Mabe notes.

“The sheer number of possible combinations in today’s sophisticated rating structure compared to prior ones, it does make it more difficult and more challenging for brokers to understand, and therefore the end consumer as well,” says Scott Lennox, vice president of pricing and research and development for The Economical. “I think that’s where they [brokers] struggle. You can’t necessarily communicate that level of complexity because there are just far too many combinations. As Kathy mentions, we can communicate key messages, the key drivers of information [i.e. the risk characteristics]. But there is always going to be a gap between what we communicate and the millions of possible combinations that gives it that little bit of uncertainty. I think that’s where the education and the communication become so critical with this.”

The other thing, too, is that brokers will become more used to it and therefore more comfortable with it, Lennox adds. Practice makes perfect, so to speak. Understanding will gradually displace uncertainty as the technique becomes more widespread in its use. “It’s something that is a little bit different, and I think that until [brokers] really work with it and understand it a little bit better from that perspective, there is going to be a little uncertainty that they have. But it does allow us to be able to compete directly with one of the brokers’ major concerns, which is the direct market.”

Insurers generally note the increasing sophistication of their underwriting techniques often benefits consumers, because the pricing more adequately reflects the actual risk. This means drivers who are good risks are not subsidizing drivers who are poor risks. The challenge will be in how to explain this to consumers.

Lennox gives one example of how sophistication might actually reduce price for a consumer. “As an example, we rate by age as well as by years licensed,” he says. “If you look at the analysis on the one-variable-at-a-time method, we call that a one-way analysis, your conclusion would be that younger operators have a worse experience than older operators, and so they should be paying maybe 20% more, for example.

“You could also do the analysis based on the number of years [the consumer was] licensed, and your analysis would show those who don’t have a lot of experience are maybe 20% worse than those who have longer experience.

“A young, inexperienced operator could be paying 1.2 times 1.2 [based on age and years licensed, respectively], so they would be paying double. The key is, what is driving the worse experience of that group? Is that they don’t have a lot of experience, or is it age-related? You want to make sure you are charging the right thing for the right reason.”

In the above example, there is an overlap – in statistical terms, a ‘correlation’ – between age and years licensed. For example, the older the driver’s age, the more yea
rs he or she will likely be licensed. Using the multivariate technique, an insurer can account for this overlap when determining the premium price. “So an individual that’s 50 years old who immediately gets their license doesn’t pay 20% more [because of the zero years licensed] because the 20% [premium increase] for the inexperienced operator maybe mostly driven by younger operators,” as Lennox explains. “So that’s a way that you remove the double-counting, in a way.”

To the average consumer, these methods likely appear to be very technical, and the different combinations of variables are bewildering, even to insurers. In some situations, a broker or insurer couldn’t possibly begin to describe all of the factors and formulae that go into the generation of a consumer’s premium.

Still, insurers do have a role to play in explaining the method to brokers and consumers. Transparency in this context, Mabe says, is about telling the customer which variables or combinations of variables made the most difference in price.

“You can say, ‘Here is the price,'” says Mabe. “‘Now it’s a little bit more than last year. We can’t tell you the whole formula that goes into it, but here are the three things that you can do to improve this price.’

“It could be a variety of different things. It could be everything from the driving record has changed, the vehicle is a year older and that takes the premium down, or it’s a new vehicle. But if you can get at that, it can help consumers.”