12,399 views

, 17 tweets, 3 min read

The effect of big data, predictive modeling & AI will have a slow arching impact on insurance. It is NOT that these techs aren't valuable. They are! It's more about the incentives and economics of the industry. Let me explain:

@Lemonade_Inc

@Lemonade_Inc

Right now, the two main KPIs in which we judge underwriting is 1. premium and 2. profits. Premium is immediate, profitability takes time to determine (for some lines...years). Even startups like @Lemonade_Inc measure themselves on these two KPIs. This is insurance economics 101

Because of this, there has been & will continue to be a lot of pressure for top-line growth. This creates a perverse incentive to grow, grow, grow that top-line. But, insurance IS a different industry than all others. Any rapid growth can have dire consequences.

Warren Buffett has said many times that insurance is a mediocre business. It is highly regulated, capital intensive, products can be easily copied and worse...no one actually enjoys buying the product because who wants to pay hundreds of $$$ for a piece of paper??

Because products are difficult to differentiate and understand, buyers struggle with differentiating competing products. When this happens, they judge the product on price. So you want to grow really fast...lower the price! Easy-peasy....but...

due to the nature of the regulations and capital nature of the business, margins are generally VERY thin. Lowering prices to achieve growth has rarely worked and more often than not, the company runs into an existential moment down the road.

The losses will eventually mount. This may force the regulators or rating agencies to change their view of the company and the death spiral begins. They raise their rates to appease their stakeholders, then lose market share, which accelerates their losses. Rinse, repeat

So big data, ML, AI and predictive models to the rescue right? Eh, not so fast. Vendors and creators of this technology like to pitch the ability to more finely segment their underwriting, separating their bad risks from good risks. Good idea except...

this means that unless the company has the ability to use this tech to mine new markets, what this technology will immediately do, if implemented, is reduce the amount of premium they take in, NOW.

Yes, this business is likely to be more profitable (maybe), but the company won't recognize that for potentially years. You know what their stakeholders will immediately discover? The reduction of premium...ruh roh!

C-level execs are more focused on their existential moment vs company's existential moment. It is easier to grow fast now and kick the profitability can down the road to the next CEO.

& a note on this tech making companies more profitable. If all companies are adopting this tech, then there will be a marginal edge to the tech & we are all sprinting just to keep pace.

We just keep moving from low margin businesses to a new era with more tech & the same low margin businesses. Insurance is one of the LEAST innovative industries out there (due to regs, capital & mediocre talent). This new tech is ONLY helpful to us, IF, we use it to design NEW...

NEW PRODUCTS
NEW BUSINESS MODELS
There is a veritable paucity of this in our industry. The "hey let's take this product with bad economics, throw some AI on it with a slick website and we'll dominate" is ignorant of the gravity sucking it into the black hole

There are MORE exposures for which we are underinsured/uninsured than for which we ARE insured. This new technology is marvelous...but the benefit of this new tech is marginal until we get the cojones to build ANEW!

So no, I am not a curmudgeon. I am a realist. I have seen the wizard behind the drapes and it is utterly unimpressive. They have no current ability to use this technology for what the world needs it for. Someday, we will. But not anytime soon