In July, I made a blog post called “Big, Bad Data” which was reprinted by Insurance Thought Leadership. Not long thereafter, someone posted a rebuttal to that article called “Setting the Record Straight on Big Data” That included this criticism:
“[T]he premise of the article showed little understanding for the depth and complexity of information sought by insurers to evaluate and price risk, and the burdens for customers and their agents to provide that information. The article also tried to simplify a complex subject into good versus bad because of specific instances of incorrect information sourced from the public domain.”
This was my response in the Comments section of that article:
Big data done accurately isn’t the issue. Big data done poorly is. For an article I was working on this week, I spent most of Monday doing online homeowners quotes. I used my own home, a former residence, and one other residence as the basis for the six quotes I got. On EVERY quote, the “public” data used was wrong. Every one. And in each case, the inaccuracy would result in underinsurance.
No doubt, as you mention in your article, there are much more sophisticated and reliable sources and techniques. My article wasn’t about those, it was about the ones that aren’t reliable and how do we or regulators differentiate between them. it was also about raising the issue of how third-party data sources can be vetted to ensure that their information is accurate and current.
As a matter of coincidence, I received the article below in Chris Burand’s latest newsletter and he gave me permission to reprint it. If you are not familiar with Chris, he is one of the very elite agency management consultants in the country, based in Colorado. I encourage you to subscribe to his monthly newsletter and you can access that from the link below. Chris is also an expert at articulating important concepts like this one involving a form of “price optimization” (a rating concept every regulator who has considered it has prohibited):
Insurance is Not a Magazine Subscription
Magazines and insurance seem to have three commonalities:
- Each depends heavily on renewals for profit.
- Each originally, in part, used the term “subscription” though only magazines commonly use this term today.
- Each wants to charge more, often far more, at renewal.
This is where the commonalities end and the last commonality should not exist. Magazines are priced at a market rate. Insurance is supposed to be priced at actuarially supported rates with only so much consideration to the market and profit because insurance is considered a public good. Insurance is heavily regulated because of its importance to citizens and commerce. Magazines just don’t have the same relevance.
A real need exists to balance company/agency profitability and public affordability so that public policy is best served. In other words, insurance is supposed to be priced so that the most people possible can afford it because more people possessing insurance is the greatest spread of risk possible resulting in the lowest overall cost, and the best societal results. It works for everyone: society, consumers, agents, and insurance carriers. This combination really goes to the heart of the insurance industry. It is somewhat egalitarian in nature though almost no consumer will ever see it that way, and maybe that is because the industry is not working the way it should.
Pricing has changed significantly and is set to change even more, and in ways completely novel to the industry. Magazine renewal pricing is an example. Insurance companies probably (actually they almost certainly) bought a study from one or more large consulting firms who concluded that companies could charge x% more on renewal without any actuarial justification. After all, why would an account become riskier at renewal, unless the company is constantly developing more information in the first year? Since increased renewal rates is widespread behavior, this suggests if data is developed the first year that indicates more rate, carriers are not asking the correct questions on the initial application and are not in a hurry to fix their applications. Otherwise, they know they can just charge more. While true they will lose some accounts at renewal when they raise rates, the net gain on the accounts that stay will outweigh the loss resulting in a net gain. Different economic terms exist for the different varieties of price sensitivity but most fall under the term “price elasticity.” Price elasticity has absolutely nothing to do with actuarially sound pricing.
Moreover, companies have identified they can keep more of these accounts if the agent gets out of the way. The agency variable is an important reason companies are pushing service centers. (A question: Why do companies need agents or, at least pay agents renewal commissions if the company does all the work while achieving a higher retention rate? Just asking a question more agents need to ask themselves.)
The net result is a magazine renewal pricing program. I completely understand and appreciate the opportunity carriers have identified and partially realized. Any executive running a company would have to choose this strategy once the data was presented. This strategy is a contributing reason why insurance companies have been so profitable the last twelve years. From a public policy perspective, I am not confident pricing insurance like magazines is in the public’s or even the industry’s best interest.
A newer pricing factor is the supposed ability to bypass the law of large numbers and price accounts with extreme individual precision (the statistical argument as to whether this strategy works must await another day but it is not a foregone conclusion such precision works). Assuming for now this hypothesis is correct, insurance will be made available to more consumers and businesses, though maybe not at affordable rates, is a given. The reason is that because within the law of large numbers, a certain unpredictability exists as to which account will have material losses. Pricing therefore charges those who do not have claims a huge premium while greatly undercharging those that will have a claim. Actuarially, on average, the premiums and discounts will average out, i.e., the beauty of the law of large numbers.
However, if pricing is precise, the best accounts’ premiums will decrease significantly, maybe by 50% or more. The worst accounts’ premiums will increase by thousands or tens of thousands of percent. If too many people are priced out of the market, the market likely will not work well which is just one reason the theory of such precise pricing may not work. Additionally, I cannot imagine how it is in the public’s best interest. Just consider this: quite a few uninsured drivers are already uninsured because they are bad drivers. This is why UM insurance is so important. What happens if uninsured drivers increase by 20% or 30%?
Another factor is how some insurance distribution disrupters have flaunted insurance regulations, regulations designed to protect the public and pricing integrity. The press has widely reported the shenanigans of an online independent agency/broker funded by private equity. Besides the normal ethical mores a company should observe, for their own good and the public’s, this one reportedly created a software program to hide from insurance commissioners their employees’ lack of insurance licenses. Insurance pricing and regulation are co-dependent. Insurance costs more when employees need licenses and licenses are another protection for the public because insurance is, again, considered a public good. Cheating by not purchasing licenses changes pricing.
The same firm has been questioned by some relative to conforming to rebating laws. Rebating is prohibited because rate filings list x% for agent commissions. Rebating arguably demonstrates that x% commission should be x% minus y% commission. An actuarial factor is not applicable and therefore, all customers should really pay x% minus y%, not just some consumers.
Anti-rebating rules are levelers. An agency can more easily afford rebates when one does not have to pay for licenses. Foregoing licenses, regardless of how easy they are to obtain, is not in the public’s best interest.
The insurance commissioners have heavy workloads and plenty on their plate of more immediacy. I know they are considering each of these factors and I am not naïve enough to suggest the industry police itself on these matters. The distribution of education and knowledge helps. Keeping what is happening quiet does not benefit anyone except the most aggressive parties. My recommendation is for all associations and regulators to consider a loud public discussion and then make the rules enforcement consistent, extremely consistent, for all.
I recommend agents keep their clients’ best interests in mind by actually working the renewals. If you want a service center, build your own. Companies do not need to pay agents a renewal commission for doing nothing on a renewal. For now, they are just being benevolent. These scenarios remind me so much of the proverb involving the frog bathing in the warm water thinking they have a free warm bath until the water is boiling and they’re dead.
I’ve spoken about “price optimization” in seminars and webinars, along with the issue of “big data” to allegedly make rating more and more accurate. The essence of insurance is the spread of risk and its cost. If “big data” magically enabled carriers to predict losses in smaller and smaller groups, at some point they won’t be able to afford it, so they’ll go without insurance, endangering the public. At the extreme, you have the Tom Cruise movie “Minority Report” where the government could predict whether someone would commit a crime and arrest them before it happened.
To what end, does this level of predictability serve society? No doubt, the motivation for carriers is reduced underwriting/pricing costs, along with improved loss experience, resulting in increased profitability. However, isn’t profitability regulated in all states? How many state insurance departments are going to allow any insurer’s profit margin to approach that of Apple or Microsoft? So, from a bottom-line standpoint, what’s the point? And, again, how does this benefit the public?
Going back to my opening in this blog post about the exchange of ideas at Insurance Thought Leadership about “big data,” I’ll refer back to the rebuttal comment, “The article also tried to simplify a complex subject into good versus bad because of specific instances of incorrect information sourced from the public domain.”
I’d say a “specific instance” of incorrect information is material if it results, for example, in someone’s home being underinsured by at least a third. As I explained in my comment to the rebuttal article, I got six online homeowners insurance quotes for three homes that I had personal knowledge of. In EVERY quote, wrong public information was used. In one instance, the home’s replacement cost that was the basis for the premium was $186,000 and I know that actual cost should be over $300,000. How many “specific instances” does it take before the magic of “big data” begins to take the form of black magic?
Another example I’ve given on many occasions involved my last three years with a homeowners insurance carrier. In the first year of this span, I was hit with a $1,000 premium increase based on a deteriorated insurance/credit score. My investigation found that the carrier had used credit information from someone else named Wilson in Colorado. In the third year of that span, the carrier attempted to increase my premium by $700, citing three “reason codes” for yet again a deterioration in my insurance score. I was able once again to show that their information was incorrect and they admitted it. What was interesting was that I discovered from several agents that these same three “reason codes” had been cited for premium increases on other accounts. In my case, the carrier’s VP of underwriting admitted there was no problem with my credit score, but the increase was justified by a statewide rate increase. When I checked with an insurance department contact, I found that this carrier had not had a rate increase in at least three years. At that point, I moved my account.
In the case of the rating factors like home square footage and credit scores, at least there is some ability to vet and challenge this information. However, when rating consists almost solely of “black box” algorithms that employ “big data” of questionable accuracy and currency, what recourse do consumers have and how can regulators possibly ensure compliance with state laws that rates and/or premiums must be adequate, not excessive, and not unfairly discriminatory?
I guess the answer is that we’ll just all have to trust the sources to be altruistic and not self-serving? What could go wrong?
Photo by Kevin Krejci