Jonathan Schwartz Chief Innovation Actuary

August 2018

The data revolution and rapid growth of the Insurtech industry has made the actuary, the professional who calculates risk in insurance, one of the most sought after individuals in the industry.
Or as the Innovation Manager of Lloyd’s of London told me once  — “You’re an Actuary? Nowadays, that makes you a Rock Star!”

I’m defiantly not a Rock Star, but I am more popular today as an actuary than I was in High School. Beyond my work at PIX,
many people ask to meet for coffee (or hummus), especially entrepreneurs who want to run by me their ideas and projects.
I would like to present some of the issues and insights that always come up as an Insurtech FAQ.
Most of the tips here are relevant to innovative insurance products, but I hope to write another post on technological services for insurance in the future.

Business model

  • The name of the game here is cash flow. In the insurance business — money comes in and with a certain probability, money at some point goes out. A business model where the expected amount of money going out isn’t less than what came in is not viable. It doesn’t matter how innovative the product is or how many buzzwords are used to explain it.
  • To put it simple, there isn’t really “stock” in insurance. If a policy is underpriced, it can be easy to get “carried away” and sell a very large number of policies at a loss. It could take a long time to understand that this has happened, and all the while, losses will continue to accumulate.

Regulation

  • Insurance is a highly regulated industry. There is good reason for this. If an insurance company goes under, there is tremendous potential for damage to the public. These companies manage a good deal of the public’s savings and pensions and the consequences of going bankrupt would be catastrophic.
  • The main goal of regulation is to protect the policy holders and not the insurance companies. It’s important to understand that this regulation with its countless rules and conditions contributes to high operational costs.
  • Regulation also demands that almost any new step an insurance company undertakes in its business such as changing the wording of a policy or pricing scales needs to be approved by the Regulator. Sometimes it’s easier to a build a product than to approve it.

Distribution channels

  • When selling insurance via a direct channel without an agent, the insurer spends funds on marketing regardless of whether the sale is made or not. However, when the insurer works with agents, the commission is a success fee, only being paid when a sale is made.
  • Agent commissions are paid on renewal as well, unlike the direct insurer’s client acquisition cost which is only incurred on the initial acquisition and not on renewal.
  • When distributing insurance products through agents, it is very hard to cross or up-sale. The client “belongs” to the agent and he can up-sell to whichever insurer he wants.

Data & Models

  • Returning to the subject of regulation, most advanced machine learning models are usually “black boxes” in which the internal “reasoning” of the model is unknown. It can very hard to approve this type of model where it’s unclear how it will behave in different scenarios.
  • A model that assesses risk more precisely won’t lower the premium for all the policy holders. It will just partition the risk in a more exact manner lowering the prices for some and raising for others.
  • Before building on the idea of “developing a more exact risk model”, one must ask himself or herself why it hasn’t been done already, bearing in mind that we are dealing with an industry that’s entire business model is based on risk assessment. Answers to this question can expose future barriers and save time and trouble later on.

Final words

Most importantly — Insurance is already exceedingly complicated, so try to keep it simple!