Insurtech and the Law of Large Numbers

I read a comment from a consumer who purchased a renter's policy from a well-known, low-price, direct carrier for $25 a month and got a quote for $5 a month from a well-known, though brand new startup. The consumer asked online of the startup, "Are you real?" One of its executives replied, "Yes, and we can charge so little because of our technology!" (I'm paraphrasing slightly for confidentiality.) That is some kind of technology to legitimately charge 80% less. If the technology is that great, everyone else should just pack up and quit now. But, first, I will go through some simple math. I'll use the incumbent carrier's results using publicly available data. The limitation is that I do not have line-item expense data down to the renter's policy level. This might make a difference, but, because the price difference is 80%, the difference is not material for this explanation. See also: Chasing the Right Numbers on Claims   The incumbent's overall expense ratio excluding loss adjustment is about 14% of written premium. The overall industry average over the last 10 years is 27.1%, and in 2016 it was 27.7%. Expense ratios tend to be stable, stubbornly so when companies need to decrease them. The incumbent's profit margin excluding investment revenue was approximately 5%. Therefore, it pays out approximately 70% of premiums in claims (total industry average for all lines is approximately 59%). This means that, if this carrier had no expenses and no need for profit, it would have to charge $.70 per dollar of premium just to  break even. The new competitor (let's call it "NewCo") is charging approximately 80% less. NewCo is too new to use its actual expenses as comparisons (the profit margin was hugely negative last year, which is normal for a startup, even one with great technology). 80% less, though, is less than the incumbent's expense ratio when loss adjustment expense is included (LAE is approximately 10%). In other words, the incumbent's expense ratio including LAE is approximately 30%, one of the lowest in the industry, and yet NewCo can justify a rate of $.20? To the best of my knowledge, no developed personal lines company has a sub-20% expense ratio including LAE. I suppose that, if everyone worked for free, if reinsurance were free and if the great technology was free, it might be possible. For NewCo's executive to be accurate, its technology must be so good that the great technology, his salary, others' salaries, auditing fees, license fees, all other expenses and all losses must be less than $.20 on the dollar. Is the executive correct? We'll have to wait and see. Giving the startup the benefit of the doubt, the only way a company can make money at 20% is if the technology identifies prospects that will not have claims, except in a highly unlikely scenario, such as maybe Black Swan events. The policy would have to be really a de facto catastrophe policy, even though the insured does not see it as such. Another possibility is that the forms are not comparable, which means NewCo's form is disingenuous or a de facto cat policy from a different angle. Based on the executive's response, though, no indication was made coverage was less, so I'm going to assume the forms are comparable. If I am wrong, a serious disclosure should have been made. I am going to extend the benefit of the doubt further. If NewCo's technology really is that good, to select people highly unlikely to have a claim, then those people do not really need insurance. They are just wasting much less at $60 per year than $300 per year. Going further into the implications for the industry: There have to be clients so unlikely to incur a claim that rates can legitimately be 80% less. The law of large numbers is based on the concept that a company cannot, within reason, predetermine which of 100,000 renters will have a theft or fire. The company can only identify the probable number of claims and the claim dollars it'll incur from these 100,000 renters collectively. If the carrier charges enough for all 100,000 policyholders (law of large numbers) but does not identify specifically who will have a claim because historically (and maybe still today) that is not predictable, then the company can make a small profit. The profit on some clients will be 70%, and on others it will be -1,000%, but, collectively, the underwriting profit will be 5%. If NewCo's executive is correct, what he was really saying is that its technology knows exactly who will have a claim--true predictive modeling down to the individual level. This means the consumers likely to have claims will pay much, much, much more. The carriers and agents stuck with these unfortunate clients will have serious problems, too, because the rates they have to charge may be so high as to be unaffordable. As much as people hate paying premiums that are always too high, historically insurance was egalitarian in many ways because all clients in a pool were treated somewhat equally. See also: How to Make Your Numbers Jump   Of course, NewCo could just be seriously underpricing its product like hundreds of new carriers that have gone before and often failed. Insurance history is littered with the tombstones of carriers that have "figured out something smarter" but were really just underpricing. It would not be the first time, and it will not be the last, where an executive thought the key advantage was technology when, in reality, the carrier did not even possess a competitive advantage. I recently heard another insurance C-suite executive advise that its advantage was technology when a review of its financials suggests it is not even investing in technology beyond some interesting reserving models. There was one other aspect of the web-based conversation between the consumer shopping renters' insurance and NewCo's executive. When asked if the company was real, the executive also advised that NewCo was real because regulators have licensed it. Please understand that, just because regulators have licensed a company, does not always mean much in the real world. I believe that almost all insurance companies that have failed were licensed. For an executive to use this as proof of "being real" causes just a smidgen of skepticism, especially when combined with a need to write to a combined ratio of 20%. The logic doesn't add up.